where are iraq’s poor: mapping poverty in iraq - world...

Where are Iraq’s Poor: Mapping Poverty in Iraq

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

wb350881

Typewritten Text

97644

Acknowledgements

This work was led by Tara Vishwanath (Lead Economist, GPVDR) with a core team comprising Dhiraj Sharma (ETC, GPVDR), Nandini Krishnan (Senior Economist, GPVDR), and Brian Blankespoor (Environment Specialist, DECCT).

We are grateful to Dr. Mehdi Al-Alak (Chair of the Poverty Reduction Strategy High Committee and Deputy Minister of Planning), Ms. Najla Ali Murad (Executive General Manager of the Poverty Reduction Strategy), Mr. Serwan Mohamed (Director, KRSO), and Mr. Qusay Raoof Abdulfatah (Liv-ing Conditions Statistics Director, CSO) for their commitment and dedication to the project. We also acknowledge the contribution on the draft report of the members of Poverty Technical High Committee of the Government of Iraq, representatives from academic institutions, the Ministry of Planning, Education and Social Affairs, and colleagues from the Central Statistics Office and the Kurdistan Region Statistics during the Beirut workshop in October 2014.

We are thankful to our peer reviewers - Kenneth Simler (Senior Economist, GPVDR) and Nobuo Yoshida (Senior Economist, GPVDR) – for their valuable comments.

Finally, we acknowledge the support of TACBF Trust Fund for financing a significant part of the work and the support and encouragement of Ferid Belhaj (Country Director, MNC02), Robert Bou Jaoude (Country Manager, MNCIQ), and Pilar Maisterra (Country Program Coordinator, MNCA2).

Table of Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2. Poverty Mapping Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Survey-to-census imputation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Survey-to-survey imputation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

3. Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

4. Candidate Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

5. Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

6 Poverty Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

7. Administrative Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

8. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Governorate level predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Robustness checks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Analysis of standard errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

9. Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

Iraq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

Duhouk, Nineveh and Erbil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Sulaimaniya and Diyala . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

Kirkuk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Anbar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

Salahaddin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

Baghdad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

Babylon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39


Kerbala . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

Wasit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Najaf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Qadisiyah . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

Muthana . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Thi Qar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

Missan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

Basrah . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

10. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

11. References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

12. Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

Appendix A: Headcount Rate and Number of Poor (by Nahiya) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

Appendix B: GLS Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

Appendix C: Summary Statistics of Key Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

List of Tables

Table 1: Poverty lines (ID per person per month) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Table 2: Comparison of direct and predicted estimates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Table 3: Ten nahiyas with the highest poverty headcount rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Table 4: Ten nahiyas with the most number of poor people . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

Table A1: Duhouk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

Table A2: Nineveh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

Table A3: Erbil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

Table A4: Sulaimaniya . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

Table A5: Diyala . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

Table A6: Kirkuk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

Table A7: Anbar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

Table A8: Salahadin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

Table A9: Baghdad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

iv

Table of Contents

v

Table A10: Babylon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

Table A11: Kerbala . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

Table A12: Wasit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

Table A13: Najaf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

Table A14: Qadisiya . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

Table A15: Muthanna . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

Table A16: Thi Qar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

Table A17: Missan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

Table A18: Basrah . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

List of Figures

Figure 1: Administrative structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Figure 2: Standard error of estimated poverty rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Figure 3: CDF of standard errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Figure 4: Headcount rate – Iraq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

Figure 5: Number of poor – Iraq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Figure 6: Headcount rate – Duhouk, Nineveh and Erbil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Figure 7: Number of poor – Duhouk, Nineveh and Erbil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

Figure 8: Headcount rate – Diyala and Sulaimaniyah . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

Figure 9: Number of poor – Diyala and Sulaimaniyah . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

Figure 10: Headcount rate – Kirkuk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Figure 11: Number of poor – Kirkuk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

Figure 12: Headcount rate – Anbar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

Figure 13: Number of poor – Anbar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

Figure 14: Headcount rate – Salahaddin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

Figure 15: Number of poor – Salahaddin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

Figure 16: Headcount rate – Baghdad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

Figure 17: Number of poor – Baghdad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

Figure 18: Headcount rate – Babylon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

Figure 19: Number of poor – Babylon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40


Figure 20: Headcount rate – Kerbela . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

Figure 21: Number of poor – Kerbala . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

Figure 22: Headcount rate – Wasit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Figure 23: Number of poor – Wasit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Figure 24: Headcount rate – Najaf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Figure 25: Number of poor – Najaf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

Figure 26: Headcount rate – Qadisiyah . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

Figure 27: Number of poor – Qadisiyah . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Figure 28: Headcount rate – Muthana . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Figure 29: Number of poor – Muthana. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

Figure 30: Headcount rate – Thi-Qar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

Figure 31: Number of poor – Thi-Qar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

Figure 32: Headcount rate – Missan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

Figure 33: Number of poor – Missan. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

Figure 34: Headcount rate – Basrah . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

Figure 35: Number of poor – Basrah . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

vi

1

1

Introduction

Measuring poverty and tracking it over time is an important prerequisite to national economic planning. Until recently, reliable information on the standard of living of all Iraqis was not forth-coming. Although household budget surveys were conducted in 1992/93 and 2002, they ex-cluded the Kurdistan region. Absence of o�cial data on household expenditure or poverty line hampered the ability of Iraqi policymakers to understand the extent of the problem, analyze their causes, and devise appropriate policies.

Iraq Household Socioeconomic Survey 2006/07 (IHSES) was the �rst survey of its kind since 1988 to cover all 18 governorates. The survey collected rich information on income, expenditure, employment, housing, education, health, and other socioeconomic indicators. It was followed by construction of an o�cial poverty line and assessment of causes and consequences of poverty. IHSES 2006/07 also formed the basis for Iraq’s National Strategy for Poverty Reduction 2009.

Building on the experience of the �rst IHSES survey and using international best practice on sampling and questionnaire design and survey implementation, the second round of IHSES was �elded in 2012/13. A nationally representative sample of more than 25,000 households from all 18 governorates was interviewed on multiple topics. A comprehensive assessment of household welfare and its determinants was completed in 2014 culminating in the report The Unful�lled Promise of Oil and Growth: Poverty, Inclusion, and Welfare in Iraq (The World Bank, 2014).

Although national, regional, governorate, and qhada (district) level poverty rates can be estimat-ed from the IHSES 2012/13 data, sampling design and sample size of the survey does not allow reliable and representative poverty estimates at the nahiya (sub-district) level. To �ll this data gap, a larger survey was designed to collect information on correlates of household welfare like demo-graphic characteristics, education, occupation, housing, and assets and estimate small-area pover-ty rates using projection methods. This report presents results from the exercise, the �rst of its kind for Iraq. Poverty mapping not only provides a visual representation of poverty at subnational levels, it also reveals pockets of poverty and islands of prosperity where they exist. This knowledge is use-ful to inform decisions on policy design and targeting of development projects and programs.

1


2

3

3

Poverty Mapping Methodology

Survey-to-census imputation

Conventionally, the discussion and motivation for small area estimation is framed around impu-tation from survey to census. It combines the strengths of household budget survey and census to estimate poverty headcount and other indicators at small geographical areas. Budget surveys collect detailed information on household expenditure which yields average income, poverty headcount rate, and other welfare and inequality measures. The surveys usually allow subnation-al estimates at the �rst administrative level such as regions, provinces, or urban and rural areas. However, the sample size of the surveys is usually too small to make reliable statements at lower administrative units like districts, sub-districts, or census tracts. On the other hand, all households are interviewed in a census which mitigates the small sample problem. The downside is that infor-mation on income or expenditure is typically not collected in the census. Small area estimation involves modeling a relationship between expenditure and individual, household, and location characteristics in the survey and imposing the relationship on the census to estimate household expenditure. Of course, for this exercise to be possible, the variables used for modeling household expenditure must be common to the survey and the census.

We follow the small area estimation methodology developed by Elbers, Lanjouw and Lanjouw (2002) (2003) (henceforth ELL). The exercise proceeds in three steps. First, the variables common to survey and census are identi�ed. These usually include demographic characteristics like age, gender, marital status, ethnicity, education, and occupation, dwelling characteristics like owner-ship and occupancy of dwelling, type of wall, roof, �oor, kitchen, toilet, sewage, garbage disposal, lighting, cooking fuel, heating, and drinking water, and ownership of assets like livestock and poultry, agricultural land, agricultural equipments, electronic equipments, furniture, and vehicles.

Identifying the common set of variables is only the �rst step; they also have to be vetted for sta-tistical comparability. For a variable to be used in modeling, the means of the variable, if not the distributions, must be statistically indistinguishable. Many variables have means that are statisti-cally di�erent from each other which makes them ineligible to be included in the model.

2


In the second step, household expenditure is regressed on the set of candidate variables as following:

(1) ln y ic( )= Xicβ +Zcγ +uic

where ln y ic( ) is the log of per capita expenditure of household i in cluster c, Xic is the vector of household and cluster charac-teristics respectively, and uic is the vector of disturbances with distribution F(0,Σ). E�cient estimates of the betas are obtained by estimating the variance-covariance matrix of the error term (Σ )and using generalized least squares (GLS) to estimate the parameters. One of the concerns in running regression (1) is whether or not household level variables can su�ciently cap-ture the intra-cluster correlation of welfare, i.e, the correlation in household welfare due to cluster-speci�c characteristics. It is recommended that cluster level information be added to the regression, either from auxiliary data sources or by creating area-level average from the larger survey and merging it with the survey.

It is assumed that the error term consists of a location-speci�c component and a household-speci�c component that are independent of each other and uncorrelated to the observable characteristics:

(2) uic = ηc + ε ic

This speci�cation allows for correlation in household expen-diture within clusters and heteroskedasticity in the household component of the error term.1

To estimate the variance-covariance matrix Σ , equation (1) is �rst estimated using Ordinary Least Squares (OLS) to obtain the residuals uic . The location component is estimated as the mean of residuals within a cluster and the household component is estimated as the overall residual minus the location compo-nent as shown in equations (3) and (4):

(3) ηc =

1nc i=1

nc

∑uic

(4) ε ic = uic − ηc( )

Heteroskedasticity in the household error component σ ε ,ic2 is

modeled using a �exible logistic function and variance of the cluster e�ect σ η ,c

2 is estimated nonparametrically allowing for heteroskedasticity in ε ic . Having estimated σ η ,c

2 and σ ε ,ic2 and

thus Σ , e�cient estimates of the betas ( βGLS ) in equation (1) and their variance-covariance matrix V β GLS( ) are obtained from the GLS procedure.

In the third step, household welfare in the larger survey is estimated by drawing a vector of betas ( �β ) from the multivari-ate normal distribution with mean �β and variance-covariance matrix V β GLS( ), a vector of location disturbance component �ηc from a distribution with mean 0 and variance σ η ,c

2 and a vector of household error component �ε ic from a normal distribution with mean 0 and variance σ ε ,ic

2 . Finally, these components are used to estimate the household expenditure for each house-hold:

(5) ln �y ic( )= Xic�β + �ηc + �ε ic

This procedure is repeated multiple times, often 100 or more, with the headcount rate and other welfare measures calculated for each round of simulation. The average over all the rounds of simulation is then reported as the point estimate and the standard error of the estimate is derived from the standard deviation of the measures. A key underlying assumption is that of stability of beta parameters, i.e., the estimated relationship

1 The location-speci�c error component is assumed to be homoskedas-tic because the number of clusters in the household surveys is often too small to allow for heteroskedasticity.

4

Poverty Mapping Methodology

5

between expenditure and household and individual character-istics do not change between the survey and the census.

There are three sources of error in the predicted welfare estimates. Idiosyncratic error is the error due to the di�erence in realized and expected consumption. Model error arises be-cause of the possible bias in the estimated betas due to model misspeci�cation. And computational error is the error inherent in obtaining the results through simulation. The smaller the area of interest, or the fewer the number of households per area, the larger is the idiosyncratic error. Thus idiosyncratic error can be reduced by imputing at higher level of aggregation. The only constraint to computation error is computational power as it declined with the number of simulations. Modeling error can be minimized by careful selection of variables, regression speci�cations, and the subgroups for which the model is esti-mated.

In a recent critique of the ELL methodology, Tarozzi and Deaton (2009) argue that it understates the standard error, and thus overstates the precision of the estimates, if there is correlation across clusters in an area. However, this is an empirical question and it is not necessarily true in every situation. Indeed, Elbers, Lanjouw, and Leite (2009) demonstrate using data from Brazil that the degree of understatement is minimal. The e�ect of correlation within and across clusters can be minimized by introducing area level means into the consumption model to capture cluster level heterogeneity.

Newer developments in the small area estimation literature extend the ELL method to obtain more e�cient and consistent estimators. Molina and Rao (2010) propose empirical best (EB) method which improves the e�ciency of point estimates by utilizing the available information on households in domains that are in both the survey and the census and Elbers and van

der Weide (2014) add a nested error structure to the EB model to obtain a more consistent estimator than ELL.

Survey-to-survey imputation

Iraq presents a unique challenge because recent census data is not available for the country. The last Iraqi census was in 1997 which excluded the three Kurdish governorates of Duhok, Sulaimaniya, and Erbil. Instead, imputation is done on a larger household survey designed explicitly to allow for small area estimation. We apply recent adaptation of the ELL methodolo-gy to predict poverty rate from one survey to another. The core methodology remains the same—build a consumption model using household expenditure survey and use it to predict con-sumption in other surveys. This is often used to predict poverty rate over time. Household expenditure surveys are conducted infrequently, usually every three to �ve years. The absence of data makes it di�cult to estimate welfare trend in the interven-ing years. Nevertheless, if information on correlates of poverty is available for those years, poverty headcount and other wel-fare indicators can be estimated using the projection method. The assumption of stability of beta parameters becomes even more important for survey-to-survey imputation because the projection is often several years forward.

In addition to the conventional sources of error, a fourth source of error—sampling error—pertains to survey-to-survey exercise because imputation is done on survey data rather than census. Sampling error is assumed to be independent of other errors. Thus standard errors are larger than what they would be had census data been used. Sampling error in the prediction data was accounted for by exporting the simulated data to Stata and obtaining robust standard errors after “survey setting” the data, i.e., specifying household weight and strata and cluster variables.


6

7

7

Data

Imputation data for poverty mapping exercise comes from Iraq Poverty Mapping and Maternal Mortality Survey (I-PMM), a nationally representative survey of more than 300,000 households, conducted close on the heels of IHSES 2012/13. The objectives of I-PMM were twofold: to esti-mate maternal mortality rate and collect reliable information on socioeconomic indicators from a large number of households.

The I-PMM data has many advantages over a regular census data because it was designed ex-plicitly to allow for small area estimation. The I-PMM questionnaire included the expected candi-date variables for modeling and imputation. Detailed information on the size, composition, and structure of the households came from the household roster and the survey asked about other usual strong predictors of welfare like education and occupation of household members, char-acteristics of dwelling, and ownership of assets. In addition, distance from and time taken to the nearest facilities captured location-level di�erences which improved the prediction power of the regression models. Literature shows that the alteration of the wording of the question, response options, and ordering of the questions may a�ect survey responses and often it is not possible to anticipate the direction and size of such bias. Thus the questions in the I-PMM survey were adapt-ed from IHSES with little modi�cation. Oftentimes, there is a lag of many years between survey and census which strains the assumption of the stability of beta parameters. This was not an issue in Iraq because I-PMM was conducted immediately after IHSES 2012/13. Finally, to the extent that di�erences in survey implementation cause di�erences in data, this is less of a concern because both the surveys were conducted by the same team of Central Statistics O�ce (CSO) and Kurdis-tan Rregion Statistics O�ce (KRSO) sta� and the two surveys were closely coordinated.

Information on household expenditure for poverty mapping comes from Iraq Household and Socioeconomic Survey (IHSES) 2012/13, a nationally representative income and expenditure survey in Iraq. Modeled after the Living Standard Measurement Surveys (LSMS), it collects detailed data on household income and expenditure, health and education, employment and job search, displacement, housing and access to services, and many other socioeconomic indicators from more 25,000 households.

3


Besides IHSES and I-PMM, two auxiliary data sources were ex-plored to augment to the regression model: Multiple Indicator Cluster Survey (MICS) 2011 and Iraq Civil War Dataset from the Empirical Survey of Con�ict (ESOC) study. MICS 2011 is a na-tionally representative household survey focusing on indicators of women’s and children’s wellbeing like marriage and birth his-tory, child mortality, maternal and newborn health, contracep-tion, HIV/AIDS, breastfeeding, early childhood development, immunization, care of illness among other indicators. Despite having a rich set of information, the MICS 2011 data could not be used due to its sample size. The survey interviewed a total of 36,580 households, implying that the sample size at qhada and nahiya levels were too small for calculation of area-level

averages. The Iraq Civil War Dataset contains rich information on violent incidents between coalition forces and insurgents, civilian casualties, spending on reconstruction projects, oil and gas reserves, and election returns. However, the data could not be used in modeling because it does not identify new districts that have been de�ned since 2003.

In absence of auxiliary data, area-level average were calcu-lated from I-PMM and merged with the IHSES data before including them in the regression model. Area-level means improve the precision of the predictions by capturing the di�erences in household standard of living due to location characteristics.

8

9

9

Candidate Variables

The pool of variables common to the two surveys is as following:

Demographic characteristics: Gender, age, birthplace, marital status, household size, number of children, adults, and elderly in the household, dependency ratio

Education: Education level of the household head, highest level of education by any household member

Occupation: Employment status, occupation, sector of employment

Housing characteristics: Type of housing unit, main construction material of wall, total area of land and dwelling, ownership and occupancy status of dwelling, source of drinking water and electricity, type of sewage and toilet

Productive and durable assets: Ownership of cooler, refrigerator, freezer, electrical generator, cooker, television, washing machine, dishwasher, water heater, heater, electric fan, air conditioner, vacuum cleaner, motorcycle, car, and PC

Location: Distance from and time taken to the nearest school, hospital, health center, bank, bus stop, market, paved road, etc.

4


10

11

11

Modeling

The level at which regression models are run must be chosen judiciously. If a single model is speci�ed for the entire country, the implicit assumption is that the parameter estimates on the regressors are the same for all regions of the country. In other words, a national model assumes that the relationship between household expenditure and household characteristics are uniform throughout the country. This may not be a tenable assumption in a country like Iraq with wide spatial heterogeneity in incidence of violence, endowment of natural resources, and robustness of factor markets. For example, returns to education are likely higher in Baghdad where formal job market is more robust than in the poorest governorates with thin labor markets. Fitting separate models by region allows the relationship between expenditure and the explanatory variables to vary and it reduces the standard error of poverty prediction due to the error in modeling. An alternative way to allow the coe�cients to vary by region is to interact the variables with regional dummy variables in the regression. This approach is �exible enough to allow di�erential relation-ships across regions and also minimizes the chances of over�tting.

In this exercise, we proceeded in a top-down fashion to decide the level at which to model the relationship, starting from a national-level regression. The national model yielded accurate pre-diction for the whole of Iraq but the model could not capture the heterogeneity across gover-norates re�ected by poor governorate level predictions compared to direct estimates from IHSES. In the next iteration, three regional models were �tted, one each for Baghdad, Kurdistan (Duhok, Sulaimaniya, and Erbil) and Rest of Iraq (14 governorates). Accurate predictions were obtained for Baghdad and the three Kurdistan governorates but the third model was still not �exible enough to capture the di�erences across 14 governorates. Next, �ve division-level models were run for Baghdad, Kurdistan, North (Nainawa, Kirkuk, and Salah ad-Din), Centre (Anbar, Diyala, Najaf, Kerbe-la, Wasit, and Babylon) and South (Qadisiya, Thi-Qar, Muthana, Maysan, and Basrah). The governor-ate level predictions from the division level regressions were not satisfactorily close to the IHSES estimates in the North, Centre, and South.

Finally, one model for each governorate was run. One concern with running multiple models is the loss in degrees of freedom and the risk of over�tting, i.e., the models are forced to explain the

5


12

noise in the data in small sample. To avoid the problem of over-�tting, researchers recommended that the sample size be no smaller than 300 for each regression (Ahmed, Dorji, Takamatsu, & Yoshida, 2014). By this criterion, governorate level regres-sions were feasible because the governorate with the smallest sample size was Kerbela with 612 households. An important implication of running governorate-level regressions is that the variables must be comparable at the level of the governorates; a variable that is statistically indistinguishable at the national level may nevertheless be di�erent for a particular governorate. Thus only those variables whose averages are statistically simi-lar within governorates are used in each model.

Appendix B presents the results for each GLS model. The vari-ables that consistently feature as signi�cant predictors of house-hold expenditure are education level and sector of employment of the household head, household size, age composition of the

household, and ownership of durable goods like washing ma-chine, water heater, and vacuum cleaner. This is consistent with general intuition about correlates of welfare and it jibes with the �ndings of the poverty assessment report as well. Nevertheless, one should be cautious in interpreting the regression coe�-cients as causal estimates. Unlike general regression analysis, the purpose of these regression models are not to explain the caus-es of household welfare and obtain parameter estimates on ex-planatory variables. The models are simply derived for accurate prediction of per capita expenditure. It is entirely possible that some important variables for causal analysis are missing from the regression either because they are not available in both the datasets or they are statistically not similar. In such situations, the parameter estimates will be biased if there is a correlation between included and omitted variables. It is even possible for a variable to be signi�cant in the opposite direction than what one would expect by intuition or theory.

13

13

Poverty Lines

The poverty line in Iraq is derived from the cost of basic need approach. It is de�ned as the level of food expenditure necessary for minimum caloric intake and non-food expenditure necessary to maintain a minimum acceptable standard of living. The o�cial poverty line is de�ned nation-ally according to the patterns and distribution in non-food consumption in the national sample. Although it uses a single bundle of goods, it adjusts for spatial prices di�erences. To account for di�erences in tastes and habits in consumption, regional poverty lines for Kurdistan, Baghdad, and Rest of Iraq were also constructed at the request of the Iraqi government. Unlike the national poverty line, regional lines are derived based on consumption patterns within the region. The national and regional poverty lines are shown in Table 1.

To be consistent with the o�cial poverty estimates which are calculated using the unique nation-al poverty line, national poverty line is used for simulation in this exercise. A larger point is that for the purpose of poverty mapping, the choice of poverty line is not crucial. The goal of poverty map is to reveal the spatial heterogeneity in standard of living. As such, relative ranking of areas rather than absolute values is more important. The choice of the poverty line has little bearing on the ranking of areas within governorates. Indeed, correlation of within-governorate rank of nahiyas calculated using regional and national poverty lines is 0.87.

6

TABLE 1: Poverty lines (ID per person per month)

Regional poverty lines

Kurdistan 142410.7

Baghdad 115934.7

Rest of Iraq 101675.9

National poverty line 105500.4


14

15

15

Administrative Structure

There are three levels of administration in Iraq: governorates, qhadas, and nahiyas. The governor-ates of Duhok, Sulaimaniya, and Erbil form the autonomous region of Kurdistan and the governor-ate of Baghdad consists of the capital city and the outlying areas. The remaining 14 governorates constitute the rest of Iraq. Each governorate is subdivided into qhadas (districts) and nahiyas (sub-districts). The total number of qhadas and nahiyas are 120 and 393 respectively but their numbers per governorate vary widely (Figure 1).

For the purpose of planning and policymaking, it would be ideal to be able to rank all communi-ties, villages, and towns. However, there is a trade-o� between the level at which one estimates poverty and the precision of the estimates; the lower the administrative unit, or the smaller the area, the less precise the poverty estimates become. Number of households per village or town in I-PMM is too low for meaningful poverty projections. Therefore, in this exercise, the lowest admin-istrative level at which poverty rates are reported is the nahiya. This is already a stretch because some nahiyas have as few as 50 households.

7


16

FIGURE 1: Administrative structure

Iraq

Governorate Qhada Nahiya

Duhok 7 26

Nainawa 10 31

Sulaimaniya 16 61

Kirkuk 4 16

Erbil 9 41

Diyala 6 21

Anbar 8 22

Baghdad 10 32

Babylon 4 16

Kerbela 3 7

Wasit 6 17

Salah Al-Deen 8 17

Najaf 3 10

Qadisiya 4 15

Muthana 4 11

Thi-Qar 5 20

Maysan 6 15

Basrah 7 15

Total 120 393

17

17

Results

Governorate level predictions

The accuracy of the models is judged by comparing the governorate level predictions with direct estimates from IHSES 2012/13. This comparison is possible because IHSES 2012/13 is representa-tive at the governorate level. Direct and estimated poverty rates, standard errors, and z-values for the di�erence in means are presented in Table 2.2 The projections are consistent with the IHSES poverty rates: the estimates for all 18 governorates fall within the 95 percent con�dence interval of the IHSES mean and the largest absolute value of z-score is 1.48, well within the usual threshold of two standard errors.

Robustness checks

According to the IHSES, national poverty rate in 2012/13 was 18.9 percent, with the lower and upper bounds of the 95 percent con�dence interval 17.9 and 19.9 percent respectively (Table 2), while the weighted average of predicted poverty rates is 18.2 percent. Thus the estimate falls well within 2 standard deviation of the IHSES rate and the absolute di�erence is only 0.7 percent.

With 18 governorates, there are 153 pairs of between-governorate comparisons of statistical sig-ni�cance. In IHSES, 119 pairs of governorates have statistically di�erent poverty rates. The imputa-tion results are remarkably consistent with the IHSES results: of the 119 pairs, 116 are statistically di�erent in the imputation data as well. In addition, 14 new pairs of governorates are statistically di�erent in the simulation, bringing the total to 130. This also shows that the loss in precision due to modeling and sampling errors is more than o�set by the gain due to larger sample size.

8

2 z-value measures the distance between the two means in standard errors:

z = IPMM – IHSES( ) / (s .e.IPMM )2 + (s .e.IHSES )2µ µ


18

TABLE 2: Comparison of direct and predicted estimates

Sample sizeDirect estimate

(National poverty line) PovMap estimate z-value

IHSES I-PMM FGT(0) s.e. 95% CI FGT(0) s.e.

Duhok 1,348 14,475 0.058 0.008 0.043 0.073 0.059 0.010 –0.364

Nainawa 1,885 27,814 0.345 0.023 0.301 0.390 0.377 0.031 –0.576

Sulaimaniya 3,292 29,216 0.020 0.003 0.014 0.026 0.014 0.004 0.157

Kirkuk 827 12,586 0.091 0.016 0.059 0.122 0.088 0.010 0.171

Erbil 1,933 22,275 0.036 0.007 0.022 0.049 0.041 0.015 0.870

Diyala 1,272 17,977 0.205 0.018 0.170 0.239 0.208 0.031 0.175

Anbar 1,718 16,055 0.154 0.022 0.111 0.196 0.187 0.022 –1.248

Baghdad 2,150 39,729 0.120 0.012 0.096 0.145 0.108 0.005 1.355

Babylon 863 16,812 0.145 0.016 0.114 0.177 0.122 0.011 1.377

Kerbela 612 8,212 0.124 0.029 0.067 0.180 0.134 0.010 0.214

Wasit 1,291 13,226 0.261 0.029 0.204 0.318 0.247 0.033 0.290

Salah al-deen 1,717 15,283 0.166 0.015 0.137 0.194 0.163 0.024 0.739

Najaf 646 9,866 0.108 0.026 0.057 0.158 0.087 0.010 0.307

Qadisiya 858 12,356 0.441 0.026 0.390 0.492 0.448 0.014 0.162

Muthanna 862 7,456 0.525 0.036 0.455 0.596 0.544 0.056 0.714

Thi-qar 1,078 16,748 0.409 0.026 0.358 0.459 0.376 0.017 1.043

Maysan 1,288 10,376 0.423 0.033 0.357 0.488 0.365 0.043 1.137

Basrah 1,506 17,022 0.149 0.018 0.114 0.184 0.129 0.007 1.479

Iraq 25,146 307,484 0.189 0.005 0.179 0.199 0.182

Results

19

Analysis of standard errors

As discussed in the methodology section, there is some un-certainty associated with the predictions. This uncertainty is re�ected in the standard error of the estimates; the higher the standard error, the lower the precision and the less one can be con�dent about the predictions. Therefore, the estimates must be read together with their standard errors. A simplistic reading of the predicted poverty rate might lead one to conclude that one area is poorer than the other while the two areas could be statistically indistinguishable.

It is not surprising that standard errors are decreasing with the sample size. On average, standard errors at the governorate, qhada, and nahiya level are 0.020, 0.032, and 0.045 respectively for sample size of approximately 16994, 2549, and 778 house-holds. Standard errors are also decreasing in the number of households within each level of administration; governorates, qhadas, and nahiyas with fewer households have higher stan-dard error in general, illustrated by the negative slope of the line of best �t in Figure 2.The largest governorate-level standard er-ror is 0.056, while the comparable �gure for qhadas and nahiyas are 0.106 and 0.120 respectively (Figure 3). Out of 120 qhadas,

FIGURE 2: Standard error of estimated poverty rate

12

10

8

6

40 0.05 0.1 0.15

Log

(Num

ber o

f H

ouse

hold

s)

Standard error

Governorate

12

10

8

6

40 0.05 0.1 0.15

Log

(Num

ber o

f H

ouse

hold

s)

Standard error

Nahiya

12

10

8

6

40 0.05 0.1 0.15

Log

(Num

ber o

f H

ouse

hold

s)

Standard error

Qhada


20

FIGURE 3: CDF of standard errors

1

11.

11.

11.

11.

11.

5

0.9

0.9

0.8

0.8

0.7

0.7

0.6

0.6

0.5

0.5

0.4

0.4

0.3

0.3

0.2

0.2

0.1

0.1

0

0

CDF

Standard error

Governorate Qhada Nahiya

less than a �fth (23) have standard error of more than 0.050 and nine nahiyas out of 393 have standard error of more than 0.10.

An alternative way to interpret standard errors is through con�dence interval. Con�dence interval is a range of values which is likely to include an unknown population parameter. The interval varies from sample to sample but if from the same population, a certain independent samples are taken repeat-edly, a known percentage of the intervals will include the unknown parameter. Almost half the qhadas (58 of 120) and a quarter of the nahiyas (90 of 393) have a con�dence interval of 10 percentage points or lower.

Overall, the estimates are remarkable precise considering the relatively small number of households per nahiya and the sampling error in the imputation data. High degree of compa-rability between IHSES and IPMM data is likely responsible for the precision in the estimates. Details on nahiya level estimates and their standard errors are presented in Appendix A.

21

21

Maps

Iraq

The visual representation of spatial distribution of poverty con�rms many of the previous analyti-cal �ndings and intuitive notions of prevalence of poverty in the country. The added value of the map is the revelation of pockets of poverty and islands of prosperity that would not have been apparent otherwise. For instance, there is considerable heterogeneity in headcount rate even within the poorest governorates. Nahiya-level headcount rate in Maysan ranges from 21 to 72 percent, a spread of 51 percentage points. The poverty rate in Thi-Qar is 37.6 percent, more than twice the national rate; nevertheless, Al-Nasiriya Qadha Center has a poverty rate of 23.1 percent, only slightly above the national average. Although the poverty rate in Baghdad is 10.8 percent, estimated poverty rate of al-Mishahda nahiya in Baghdad is 49.5 percent, almost �ve times the governorate average, and it is statistically di�erent from all other nahiyas in the governorate. Only through the small area estimation exercise do these patterns become apparent.

Headcount rate and number of poor are alternative ways to visualize the spatial pattern of pover-ty. Areas with high headcount rate are typically areas that have historically been marginalized and left out of the development process. Where equity consideration and national political preference prioritizes the development of such regions, it is justi�ed to allocate disproportionately higher budget to those areas to increase linkage with markets and build infrastructure networks like roads, electricity grid, and irrigation canals. Such areas are usually remote, rural, and sparsely pop-ulated. However, if the national priority is to reduce poverty headcount, it is e�cient to focus on where most of the poor live. Usually, despite low headcount rate, urban and semi-urban areas are host to more poor people than the poorest parts because of the population size. This has import-ant policy implications as we discuss in the concluding section.

Often there is little overlap between areas with the highest poverty rate and areas with the most number of poor people (Figures 4 and 5 and Tables 3 and 4). Not surprisingly, nahiyas with the highest poverty rate are in the Southern governorates of Maysan, Muthana, and Qadisiya. The nahiya with the highest headcount rate is Gammas in Qadisiya (77 percent) followed closely by

9


22

FIGURE 4: Headcount rate – Iraq

Maps

23

FIGURE 5: Number of poor – Iraq


24

has 1.3 million residents. Nahiyas like Baghdad al-Jedeeda and Al-Basrah Qadha Center are further examples of areas that host a large number of poor despite having a low headcount rate. A comparison of Figures 4 and 5 also make this apparent: nahiyas shaded in deep red in the �rst map have lighter hues in the second and vice versa.

Al-Hilal in Muthana (73 percent) and Bani Hashim in Maysan (72 percent). On the other hand, nahiyas with the highest number of poor are urban centers with many residents. The nahiya with the most number of poor people is Al-Mosal Qadha Center where the city of Mosul is located. Although the predicted headcount rate in Mosul is 31 percent, far lower that some other nahiyas in Nainawa, number of poor is high because it

TABLE 3: Ten nahiyas with the highest poverty headcount rates

Governorate Nahiya PopulationHeadcount

rate Std. Err. [95% Con�dence Interval]Number of

poor

Maysan Said Ahmed Al-Rifaai Nahia 10450 0.67 0.08 0.50 0.82 6968

Maysan Qalat Saleh Qadha Center 53955 0.68 0.06 0.55 0.77 36808

Maysan Al-Ezair Nahia 29933 0.70 0.05 0.58 0.79 20875

Muthana Al-Daraji Nahia 18258 0.70 0.09 0.51 0.86 12740

Muthana Al-Sowair Nahia 39066 0.70 0.09 0.51 0.87 27428

Muthana Al-Warka Nahia 93216 0.71 0.09 0.52 0.88 65988

Muthana Al-Najmi Nahia 32461 0.72 0.08 0.54 0.88 23453

Maysan Bani Hashim Nahia 20003 0.72 0.06 0.60 0.83 14498

Muthana Al-Hilal Nahia 31800 0.73 0.08 0.55 0.88 23255

Qadisiya Gammas Nahia 87526 0.77 0.04 0.68 0.83 67351

Maps

25

TABLE 4: Ten nahiyas with the most number of poor people

Governorate Nahiya PopulationHeadcount

rate Std. Err. [95% Con�dence Interval]Number of

poor

Nainawa Al-Shamal Nahia 141128 0.62 0.06 0.49 0.75 87852

Nainawa Telafar Qadha Center 193602 0.49 0.08 0.32 0.66 95078

Qadisiya Al-Diwaniya Qadha Center 388415 0.27 0.04 0.20 0.35 105998

Wasit Al-Kut Qadha Center 407511 0.26 0.06 0.15 0.38 107746

Muthana Al-Samawa Qadha Center 271285 0.41 0.10 0.19 0.62 111254

Basrah Al-Basrah Qadha Center 1172742 0.10 0.02 0.06 0.13 111880

Thi-Qar Al-Nasiriya Qadha Center 498661 0.23 0.04 0.14 0.32 115141

Baghdad Baghdad Al-Jedeeda Nahia 1140276 0.11 0.02 0.07 0.15 126000

Maysan Al-Amara Qadha Center 529251 0.25 0.08 0.09 0.41 131625

Nainawa Al-Mosal Qadha Center 1289229 0.31 0.07 0.17 0.46 405591


26

Duhouk, Nineveh and Erbil

Duhouk lies in the northwest of Iraq and it is the north-ern-most governorate, bordering Turkey and Syria and is part of the Kurdistan region. Duhouk was home to many internally displaced persons (IDPs) even before the in�ux

of Syrian refugees in 2014 and the Da’ash related internal displacement.

Overall, poverty rates in Duhouk are low at roughly 6 percent. Estimated poverty headcount rates at the nahiya level range from 3 percent in Zawait, Al-Amadia qadha center, and Sarsank

FIGURE 6: Headcount rate – Duhouk, Nineveh and Erbil

Maps

27

nahiya to 10 percent or more in Bateel, Qasrouk, Kalak and Dar-to nahiyas. However, the largest number of poor persons are in two nahiyas—Duhouk qadha center (10,600) and Zakhow qadha center (11,600)3. These nahiyas are the largest in terms of population in Duhouk; while the former is the governorate capital, the latter is a major transit and check point with Turkey.

Nineveh or Mosul governorate in northern Iraq shares its west-ern border with Syria. Its capital, Mosul, is one of the largest cities in the country. The governorate continued to experience

FIGURE 7: Number of poor – Duhouk, Nineveh and Erbil

3 The number of poor is rounded to the nearest hundred in the text.


28

violence between 2007 and 2012, and is one of the �ve poorest governorates in the country. Nahiya level estimates of pov-erty reveal a striking variation in headcount rates—from less than 15 percent in Kandinawa, Faidah and Al-Shaikhan qadha center, to 50 percent or more in Zummar, Rabia, Al-Ayadiya, Al-Qairawan, Al-Qahtanya, Al-Baaj, and Al-Shamal nahiyas. All the nahiyas with headcount rate of 50 percent or more either border Syria or lie close to the Syrian border. Many of these are also very populous nahiyas, together accounting for 27 percent of Nineveh’s poor, and many of them are home to more than 35,000 poor persons each. However, the largest number of poor people are in Mosul qadha center (406,000 persons), the capital of the governorate, which has a population of 1.28

million. Nineveh has the most disproportionate share of poor in the country—while 10 percent of the Iraqi population resides in the governorate, about 20 percent of the poor lives there.

Erbil or Hawler governorate is part of the Kurdistan region of Iraq, and has overall poverty rates of 4.1 percent. At the nahiya level, estimated poverty is almost negligible in Ainkawa, Kuwaisinjaq qadha center, Shaqlawa qadha center, andArbil qadha center at 3 percent or less. The highest poverty rate is in Siakan nahiya, where 11 percent of the population lives below the poverty line. But by far the highest number of poor are in Arbil qadha center, almost 20,000 poor persons, as compared to roughly 1,400 in Siakan, the nahiya with the highest headcount rate.

Maps

29

Sulaimaniya and Diyala

Sulaimaniya is a northern governorate of Iraq, part of the Kurdistan region, and has the lowest poverty rates in the entire country. In more than two-�fth of its nahiyas, estimated pov-erty rates are 2 percent or lower; and all but two nahiyas have headcount rate of less than ten percent. The nahiya with the

highest number of poor persons is Sulaimaniya qadha center; with an estimated poverty rate of 0.5 percent and a population of approximately 600,000 it has around 3,000 poor persons.

Diyala province lies south of Sulaimaniya and north east of Baghdad, and also borders Iran to its east. Along with Anbar and Nineveh, it continued to experience violence between

FIGURE 8: Headcount rate – Diyala and Sulaimaniyah


30

2007 and 2012. The nahiya with the least poverty rate in Diyala is Khanaqin qadha cente which has a headcount rate of ap-proximately 10 percent. Baquba nahiya, where the governorate capital is located, has the second lowest headcount rate in Diyala—15 percent—but accounts for 38,000 poor persons be-cause of its population of 260,000. Five nahiyas have headcount

rates above 25 percent, with the highest rate of 32 percent in Mendili nahiya. Besides Baquba, four other populous nahiyas account for a large number of poor persons in Diyala—Al-Muq-dadiya (34,000), Baladrooz (21,000), Al-Khalis qadha center (26,000), and Beni Saad (38,000).

FIGURE 9: Number of poor – Diyala and Sulaimaniyah

Maps

31

Kirkuk

Kirkuk governorate is located east of Nineveh and south west of the Kurdistan region. More than a third of the nahiyas in Kirkuk have poverty rates less than 10 percent including the

capital Kirkuk qadha center. About a quarter of the nahiyas have estimated headcount rate higher than 20 percent, all of whom are on the border with Salahadin. Headcount rates in-crease gradually from single digits to teen and twenties as one moves from north east to south west away from the Kurdistan

FIGURE 10: Headcount rate – Kirkuk


32

border. Despite very low headcount rates of 5 percent, the nahiya with the largest number of poor people is Kirkuk qadha center which is home to almost 900,000 people and almost

48,000 poor persons. In contrast, the nahiya with the highest headcount rate—Al-rashad—has only slightly more than 6,000 poor people because it has only 23,000 residents.

FIGURE 11: Number of poor – Kirkuk

Maps

33

Anbar

Anbar is Iraq’s largest province in terms of geographic area and shares borders with Syria, Jordan and Saudi Arabia. However, it is also the most sparsely populated. Poverty estimates vary widely across the governorate. Al-Garma and Al-Wa�a nahiyas,

for instance, which border Baghdad and Kerbala respectively, have poverty rates as high as 46 and 48 percent while Rawa qadha center has a poverty rate as low as 6 percent. Moreover, poverty rates are not negligible in the 4 largest nahiyas—Al Ramadi (15 percent), Al Habbaniya (16 percent), Al Falluja (17 percent) and Al-Garma (48 percent)—implying that these four

FIGURE 12: Headcount rate – Anbar


34

alone account for 63 percent of Anbar’s poor, with Al-Garma, Al-Fallujah, and Al-Ramadi each accounting for at least 16 per-

cent of all the poor in Anbar.

FIGURE 13: Number of poor – Anbar

Maps

35

Salahaddin

Salahadin governorate lies north of Baghdad, bordered by Anbar on the west and Diyala on the east. Estimates poverty rate in Salahaddin is 16.3 percent, but outside of Kurdistan, it is

one of the governorates with the lowest variance in headcount rate across nahiyas. The nahiya with the highest headcount rate—Sulaiman Baig nahiya—has a headcount rate of 21 percent while Al-Daur qadha center, the nahiya with the lowest headcount rate, has 9 percent. The two most populous nahiyas

FIGURE 14: Headcount rate – Salahaddin


36

are Samarra (182,000) and al-Shirqat (200,000) with poverty rates of 14 and 18 percent and 25,000 and 35,000 poor persons respectively. Together, the two nahiyas account for more than a quarter of all the poor in Salahaddin.

FIGURE 15: Number of poor – Salahaddin

Maps

37

Baghdad

Baghdad is Iraq’s smallest governorate in terms of geographical area but it is also one of its most populous. The governorate alone accounts for one-�fth of Iraq’s population and 12 per-

cent of Iraq’s poor. Its 32 nahiyas show stark di�erences in welfare levels. The �ve least poor nahiyas—Palestine, Al-Ad-hamia, Al-Karkh, al Mansour, and al-Karrda al-Sharqia—have estimated rates of poverty of less than 3 percent. The poorest nahiyas—al-Mishdada (near Fallujah)—on the other hand,

FIGURE 16: Headcount rate – Baghdad


38

has a very high poverty rate of 49 percent. The largest nahiya in terms of population, Baghdad al-Jadeeda, with 1.14 million people, and an estimated poverty rate of 11 percent, alone

accounts for almost 126,000 poor persons, the highest in Bagh-dad, or 15 percent of the poor in the governorate.

FIGURE 17: Number of poor – Baghdad

Maps

39

Babylon

Babylon province in central Iraq is located just south of Bagh-dad. Overall, poverty rates in the governorate are low (12.2 percent); eleven of sixteen nahiyas have poverty rate below

the national average. Al-Hilla qadha center, the capital of the governorate with a population of 540,000, has the lowest estimated poverty rate in the governorate of 6 percent, but ac-counts for the largest number of poor people—around 35,000 persons. At the other end, al-Talea’a nahiya, bordering Qadisiya

FIGURE 18: Headcount rate – Babylon


40

governorate, has a relatively high poverty rate with almost a quarter of its population, or more than 9,000 people, living below the poverty line. Three nahiyas with the most number of poor—Al-Qasim, Al-Ki�, and Al-Hilla—together account for

more than two-�fths of the poor in the governorate. Because of the low poverty incidence, Babylon accounts for 4.5 percent of the poor in Iraq although it is home to 6.9 percent of the population.

FIGURE 19: Number of poor – Babylon

Maps

41

Kerbala

Kerbala governorate, located east of Anbar, and west of Najaf, lies in central Iraq. It is one of Iraq’s smallest and least populated provinces (after Missan and Muthanna in the south). Between

2007 and 2012, it witnessed a signi�cant decline in poverty rates, and overall poverty rate is now below the national aver-age. Unlike many of the other governorates, there is relatively low variation in estimated poverty rates across nahiyas in the governorate, ranging from 9 percent in al-Hassaniya nahiya to

FIGURE 20: Headcount rate – Kerbela


42

20 percent in al-Hur nahiya. The most populous nahiya, Kerbala qadha center (480,000), accounts for the largest number of poor persons, around 56,000 people which is about 38 percent of the poor in Kerbala. The share of poor persons in Kerbela is

on par with its population: 3.2 percent of Iraqis and 2.3 percent of the poor reside in the governorate.

FIGURE 21: Number of poor – Kerbala

Maps

43

Wasit

Wasit governorate shares its western border with Babylon prov-ince, and its eastern border with Iran. Thirteen of seventeen nahiyas in the governorate experience poverty rates above

the national average. Even the capital, al Kut qadha center, has a quarter of its population below the poverty line, or 108,000 poor persons. The highest rates of poverty are in the eastern nahiyas bordering Iraq—Wasit (43 percent) and Sheikh Saad (48 percent)—where almost half the population is poor, but

FIGURE 22: Headcount rate – Wasit


44

because of their small population sizes, they account for only 35,000 poor persons. Overall, Wasit has 3.6 percent of the total Iraqi population and 4.8 percent of the poor.

FIGURE 23: Number of poor – Wasit

Maps

45

Najaf

Najaf governorate, south of Kerbala, also experienced rapid de-clines in poverty between the 2007 and 2012 period, and now has one of the lowest poverty rates in the country. Within the

governorate, there is little variation in estimated poverty rates, with the lowest rates of 6 percent in Najaf qadha center, and the highest of 21 percent in al-Qadisiya, with seven out of ten nahiyas having poverty rate less than 15 percent. Because of its high population share, Najaf qadha center, the capital, alone

FIGURE 24: Headcount rate – Najaf


46

accounts for 38 percent of the governorate’s poor. Al-Shabaka nahiya is one of the least populous in the country, with much of nahiya being part of the Syrian desert, and an estimated population of only 356 persons of which 57 are estimated to be

poor (16 percent). Najaf is home to 3.9 percent of the popula-tion and 1.8 percent of the poor.

FIGURE 25: Number of poor – Najaf

Maps

47

Qadisiyah

Qadisiya governorate, in south-central Iraq, was until 1976 part of the Diwaniyah governorate along with Muthanna and Najaf. It is one of the �ve poorest governorates in Iraq. Even the best-

o� nahiya, al-Diwaniya qadha center, has more than a quarter of its population living below the poverty line, almost 106,000 persons. Eight of �fteen nahiyas in the governorate have esti-mated poverty rates of 50 percent or more. By far the poorest is Gammas nahiya, with a headcount rate of 77 percent, which

FIGURE 26: Headcount rate – Qadisiyah


48

is also the poorest nahiya in the country. Indeed, 34 percent of Qadisiya’s poor live in its �ve poorest nahiyas—Gammas, Al-Shi-na�ya, Al-Salahiya, Al-Badair and Al-Sadeer—and al-Diwaniya qadha center alone accounts for a �fth of the poor. High

poverty incidence is re�ected in the disproportionate number of poor in the governorate; while 3.4 percent of Iraqis resides in Qadisiyah, 8 percent of the poor lives here.

FIGURE 27: Number of poor – Qadisiyah

Maps

49

Muthana

Muthanna governorate is located in southern Iraq and borders Saudi Arabia and Kuwait on the west and south. It is Iraq’s poor-est governorate, with even the best-o� nahiya—al-Samawa qa-

dha center—experiencing poverty rate of 41 percent. Nine out of eleven nahiyas in Muthanna have estimated headcount rate of more than 50 percent. The poorest three nahiyas—al-Warka, al-Najmi, and al-Hilal—with poverty rates of 71, 72 and 73 per-cent respectively, together account for 113,000 poor persons.

FIGURE 28: Headcount rate – Muthana


50

However, almost as many poor persons live in a single nahiya, al-Samawa qadha center, which is the most populous in the governorate. The proportion of poor Iraqis in Muthana is more

than thrice that of the proportion of Iraqis in general, 6.4 per-cent versus 2.1 percent.

FIGURE 29: Number of poor – Muthana

Maps

51

Thi Qar

Thi Qar governorate in southern Iraq lies north of Basra and east of Muthanna. It is one of Iraq’s poorest governorates, although there is signi�cant variation in nahiya level estimates

of poverty. For instance, the capital and most populous nahiya, al-Nasiriya qadha center, has a poverty rate of 23 percent. One-�fth of the nahiyas in the governorate have headcount rates above 50 percent and three-�fth of the nahiyas have poverty rates of more than 40 percent. The two poorest nahiyas, al-

FIGURE 30: Headcount rate – Thi-Qar


52

Dawaya and Said Dekhil, have more than three-�fth of their population living below the poverty line.

Al-Nasiriya qadha center has the lowest headcount rate but highest number of poor people (115,000) because of its

population size (499,000). This is apparent in Figures 30 and 31: Al-Nasiriya is shaded in “lightest” color in the headcount rate map while it has the “darkest” hue in the map for number of poor. Eleven percent of the country’s poor lives in Thi Qar while it has only 5.5 percent of the country’s population.

FIGURE 31: Number of poor – Thi-Qar

Maps

53

Missan

Missan lies in the southeast of Iraq, bordering Iran to the east, and Wasit, Thi-Qar, and Basrah governorates to the north, west, and south respectively. More than a third of Al Ahwar, or Mes-

opotamian Marshes, of south Iraq falls in Missan governorate. However, the loss of wetlands has led to deserti�cation of arable lands and uprooted many Marsh Arabs from their traditional lands and livelihoods. Despite having rich oil reserve �elds, Mis-san has one of the highest poverty rates in Iraq. The nahiya with

FIGURE 32: Headcount rate – Missan


54

the least headcount rate in Missan is Ali Al-Sharqi (21 percent) whose poverty rate is higher than the national average. More than a third of the nahiyas have headcount rate of more than 60 percent—Al-Khayr (64 percent), Al-Salam (66 percent), Said Ahmed Al-Rifaai (67 percent), Qalat Saleh qadha center (68 per-

cent), Al-Ezair (70 percent), and Bani Hashim (72 percent)—with the two poorest nahiyas bordering Iran. The areas with the most number of poor are all urban centers—Al-Maimouna qadha center (25,000), Al-Mejar Al-Kabir qadha center (32,000), Qalat Saleh qadha center (37,000), and Al-Amara qadha center

FIGURE 33: Number of poor – Missan

Maps

55

(132,000)—and together they account for 62 percent of the poor in Missan. The nahiya with the fewest number of poor is Ali Al-Sharqi, a product of its low headcount rate (21 percent) and

small population size (18,000). Missan is another governorate that has a disproportionate share of the poor: it is home to 2.9 percent of Iraqis but 5.8 percent of the poor.


56

Basrah

Basrah is Iraq’s southernmost governorate, home to Iraq’s only seaports, and borders Kuwait to the south and Iran to the east. Unlike the other southern governorates, Basra expe-

rienced signi�cant improvements in welfare between 2007 and 2012. Um Qasr nahiya, which is also the location of Um Qasr port, Iraq’s only deep water port, has poverty rates as low as 5 percent, and only 2800 poor persons. The three poorest nahiyas—al-Nashwa (25 percent), al-Dair (26 percent), and al-

FIGURE 34: Headcount rate – Basrah

Maps

57

Thagar (28 percent)—are all on the interior and border Iran. The largest number of poor people, almost 112,000 persons, live in the capital of the governorate, al-Basra qadha center which is also one of the largest cities in Iraq with a population of 1.2

million. The relative prosperity of Basrah is re�ected in less than proportionate share of the poor; Basrah is home to 5.3 percent of the poor and 7.6 percent of Iraqis.

FIGURE 35: Number of poor – Basrah


58

59

59

Conclusion

Poverty map is a tool that combines the strengths of household budget survey and population and housing census to estimate poverty rate at small level of geographical disaggregation. Na-tionally representative budget surveys collect detailed information on household consumption which allows one to estimate poverty rate up to regional and rural/urban levels. On the other hand, census interviews all households but collects information on only a limited set of variables. Small area estimation entails estimating a relationship between household expenditure and observable household characteristics using the budget survey and imposing the relationship in the census to calculate predicted consumption—and poverty status—of each household. The information is then used to estimate average poverty rate at various levels of geographical aggregation. The exercise realizes its full potential when the estimated small area poverty rates are illustrated on a color-coded map. The visual representation of spatial distribution of poverty draws more attention and interest than presenting the information on a table ever could. It can galva-nize political will and support for poverty reduction.

The discussion of poverty alleviation touches upon an important issue discussed earlier—the dis-tinction between poor areas and areas where the poor live. Areas that have historically been left out of the mainstream development process exhibit entrenched and widespread poverty. How-ever, areas with the highest poverty rate are not necessarily where most of the poor live because such areas are often sparsely populated. Urban centers and other populated areas might have low poverty rate but may host a large number of poor people due to their population size. The design and placement of programs and policies depend on whether the national priority is to target poor areas or poor people. To facilitate this discussion, the current exercise presents both poverty rate and number of poor at the nahiya level. As one can see from the maps, the areas with the highest poverty rate in Iraq are not always where most of the poor live. Indeed, there is no nahiya in common between ten poorest nahiyas and ten nahiyas with the most number of poor.

Beyond visual illustration of distribution of poverty, the map draws its power because of its useful-ness to planners and policy makers. The most obvious use of poverty maps is for the targeting of antipoverty programs. Social protection programs often collect information on proxies of house-

10


60

hold welfare to identify poor households. This is to ensure that the funds reach the poorest and that there is minimal leakage to nonpoor and undercoverage of poor. Poverty ranking of small areas can be used to re�ne the targeting mechanism and improve its e�ciency. Geographical targeting can be the “�rst stage” targeting where a program �rst identi�es the areas with the highest poverty rate before collecting detailed informa-tion for household level targeting only in those areas. In areas with high poverty rate and low population density, small area estimation may obviate the need for household level targeting because identifying the poverty status of individual households may be extremely ine�cient. Poverty ranking can also help in prioritization of placement and rollout of a program with limit-ed budget; areas with the highest poverty rate can be the �rst to receive the program or the last to lose it.

The value of poverty map is further boosted when it is combined with other geocoded information like road, electricity, and irrigation networks, schools, health centers, and access to market. Superimposing auxiliary information on poverty map can shed light on the correlates of poverty. For instance, areas with high poverty rate may have poor supply of public services, weak infrastructure network, low access to markets and other opportunities for commerce and mobility. This lets policy makers deliver interventions tailored to �t local needs.

The general caution about correlation not being equal to causation is applicable here as well; while the maps show correlates of poverty, they do not identify its causes. For exam-ple, an area may be poor because it does not have extensive road network, but the direction of causality could run in the opposite direction: a poor area with weak tax base may not have the �scal space to construct roads. The map is also not suited to analyze the impact of an intervention. Robust impact evaluations that provide credible answer to the counterfactual question “What would have been the impact in absence of the

program?” need to be designed for the purpose. Again, this is outside the scope of small area estimation method.

Besides targeting and budget allocation, poverty maps have had a broader impact as well. By presenting disaggregated in-formation on poverty in a highly accessible manner, the maps have fostered local debate on the determinants of poverty and consequences of policy. The objective poverty ranking of areas have provided countries the necessary empirical evidence to reform existing resource allocation mechanisms and placement of antipoverty programs. In countries where the maps are up-dated regularly, it has been possible to evaluate the changes in distribution of poverty over time. The maps have also served as focal instrument to coordinate the activities of multiple agen-cies. Please refer to Box 1 for international examples on the use of poverty maps.

It should be acknowledged that the numbers reported here are estimates derived from a modeling and imputation process. While great care has been taken to ensure that the estimates are as accurate as possible, including harmonization of sur-vey questions and careful selection of explanatory variables, each prediction comes with a standard error. Therefore, the uncertainty associated with the estimates must be taken into consideration while interpreting the results.

One caveat does uniquely apply to the current work. The bud-get and household surveys are from 2012 and 2013 respec-tively, hence the picture presented here represents the state of Iraq as of 2013. The country has gone through many changes since then, and especially since summer of 2014, when Islamic State militants swept through a large swath of the country. This does not negate the value of the map because it still provides a valuable retrospective snapshot of the spatial distribution of poverty. If there is a correlation between violence and poverty and inequity, it sheds light on possible causes or consequenc-es of violence to the extent violence restricts people’s ability

Conclusion

61

Poverty maps have been developed for at least 45 countries in the last �fteen years. Once a map has been developed for a country, it can be put to many uses as the following examples demonstrate.

Targeting � Bulgaria: “In Bulgaria, poverty maps account for one of �ve formal criteria used in allocating social

infrastructure projects among municipalities…The steering committee of the Social Investment Fund has found that the maps are important in the allocation process because they have helped guaran-tee an objective ranking among municipality applicants. Indeed, the committee has considered the maps so helpful that it has now integrated the small area poverty estimates into the fund’s manage-ment information system.”

� Cambodia: “In Cambodia, the World Food Programme has integrated several maps, including information on infrastructure and vulnerability to �ood and drought, into a GIS, along with small area poverty and nutrition maps. It has used the combined information to identify potential areas for its programs. The maps have also been used for resource targeting by, for example, the Ministry of Agriculture, Forestry, and Fisheries.

� Kenya: “In Kenya, the allocation formula used in the Constituency Development Fund has been revised so that 25 percent of the allocations are based on the incidence of poverty, and those areas showing higher poverty incidence receive more resources from this portion of the allocations.

Identi�cation of correlates of poverty � Sri Lanka: “Overlaying the poverty map and a map depicting access to nearby markets or cities has

demonstrated that poverty incidence is highly correlated with geographical isolation as measured by distance to the nearest market or city. This has prompted a shift to an emphasis on reaching areas that are more isolated. A similar exercise has been conducted with GIS data on drought patterns.”

Validation of existing targeting mechanism � Morocco: “In Morocco, an analysis of public expenditure and poverty has provided a measure of the

extent to which program allocations have matched the patterns of poverty (the targeting di�erential approach). Morocco has found a strong correlation between poverty and other local data. This has enabled a deeper understanding of local conditions, the evolution of social conditions, and the e�ec-tiveness of government programs in reaching poor areas.

International Experience on Uses of Poverty Maps

BOX

1(continued on next page)


62

� Vietnam: “The poverty map was overlaid with information on communes receiving funds through Program 135. The results of this exercise validated the program’s targeting criteria by showing that most communes bene�ting from Program 135 were in poor areas and that most poor areas were included in the program, although the analysis did reveal a few gaps in coverage in the Northwest region that needed attention.

Reform of existing programs � Sri Lanka: “In Sri Lanka, the small area estimates on poverty at the Divisional Secretariat level were

compared to the coverage of the Samurdhi transfer program, the largest welfare program in Sri Lanka. Only a weak correlation was found between the areas targeted by the program and the areas ranked as the poorest in the poverty map. This helped quantify the extent of mistargeting in the Samurdhi program with regard to both undercoverage and leakage. As a result, the formulas for the allocation of funds in the program were modi�ed. This was very sensitive politically, as many people stood to receive reduced bene�ts or none at all because of the changes in the allocation criteria. As a compromise, allocations remained fairly constant for existing recipient areas, but the poorest of these areas saw an increase in funding.”

Monitoring poverty over time � Ecuador: “Ecuador was one of the �rst countries to construct a series (or panel) of poverty maps.

It used data from 1990 and 2001, and the two maps helped identify areas where there had been a signi�cant increase in poverty over that time (for example, urban areas in the Coast Region, where the 1990 poverty rates were lower), as well as areas where poverty had remained largely unchanged (such as rural areas in the Coast Region, where the 1990 poverty rates were higher).”

Inter-agency coordination � Mexico: “…the president of Mexico has developed a plan to reduce poverty and promote human

development by focusing on the 50 municipalities with the highest poverty rates and the lowest hu-man development indexes. Seven ministries operating 12 di�erent but related programs now focus as a priority on the poor in these 50 municipal areas. These seven ministries have had to coordinate the 12 programs to meet the targets set out in the plan. Previously, each ministry and program had its own priorities and objectives that were implemented in isolation.


BOX

1

(continued on next page)

(continued)

Conclusion

63

� South Africa: “Data and maps on poverty, sanitation, clean water, and the incidence of cholera were used to help contain the spread of cholera in KwaZulu-Natal Province in South Africa in January 2001. Poverty and cholera data sets showed that the cholera outbreak had followed a river �oodplain and was moving through poor areas toward other poor areas. The use of the data sets assisted in produc-ing a swift, well-coordinated response by national and local government departments (health, water, and so on)...”

Sparking conversation on poverty � Morocco: “Although there is less poverty in Morocco than in most countries in Africa, the poverty

map in Morocco has highlighted the problem of persistent poverty and sparked a national conver-sation on poverty. King Mohammed VI has taken an especially keen interest, and the poverty maps have been used to help design and allocate the budget for his signature program, the National Human Development Initiative.

� Vietnam: “In Vietnam, poverty maps have revealed high levels of inequality both across and within regions. This strong message has resonated with many users and provided empirical evidence of patterns that were only suspected, but never documented.

Source: More than a Pretty Picture (Bedi, Coudouel, & Simler, 2007).


BOX

1to move in search of opportunities, discourage investment, and hinder mutually bene�cial transactions due to distrust in contract enforcement. The map is likely a more faithful repre-sentation of the current situation in the Southern governorates because they have been relatively insulated from the recent spate of violence that has convulsed the rest of the country.

It also points to the need for collecting detailed data at high frequency to be able to understand the changing circumstanc-es and respond to it appropriately. For best e�ect, the map must be updated frequently so that there is timely information on the trends of poverty and other socioeconomic indicators. New technology like mobile phone surveys and computer

assisted telephone interview (CAPI) can be used to reduce security risk and overcome logistical challenges associated with frequent surveys.

The Government of Iraq’s e�ort to measure poverty and understand its causes through regular household expenditure surveys and comprehensive poverty assessment reports is highly commendable. It is also committed to understanding the changing poverty landscape by conducting expenditure surveys at high frequency and using advanced imputation methods to estimate poverty. This will help the government be prepared and devise appropriate response in crisis and emer-gency situations.

(continued)


64

65

65

References

Ahmed, F., Dorji, C., Takamatsu, S., & Yoshida, N. (2014). Hybrid Survey to Improve the Reliability of Poverty Statistics in a Cost-E�ective Manner. The World Bank.

Bedi, T., Coudouel, A., & Simler, K. (2007). More than a Pretty Picture: Using Poverty Maps to Design Better Policies and Interventions. Washington DC: The World Bank.

Elbers, C., & Weide, R. v. (2014). Estimation of Normal Mixtures in a Nested Error Model with an Appli-cation to Small Area Estimation of Poverty and Inequality.

Elbers, C., Lanjouw, J. O., & Lanjouw, P. (2002). Micro-Level Estimation of Welfare. The World Bank.

Elbers, C., Lanjouw, J. O., & Lanjouw, P. (2003). Micro-Level Estimation of Poverty and Inequality. Econometrica, 355–364.

Elbers, C., Lanjouw, P., & Leite, P. G. (2009). Brazil within Brazil: Testing the Poverty Map Methodology in Minas Gerais. The World Bank.

Molina, I., & Rao, J. (2010). Small area estimation of poverty indicators. The Canadian Journal of Statistics, 369–385.

Tarozzi, A., & Deaton, A. (2009). Using Census and Survey Data to Estimate Poverty and Inequality for Small Areas. The Review of Economics and Statistics, 773–792.

The World Bank. (2014). The Unful�lled Promise of Oil and Growth: Poverty, Inclusion, and Welfare in Iraq 2007–2012. Washington DC: The World Bank.

11


66

67

67

12 Appendices


68

TABLE A1: Duhouk

Nahiya Population Headcount rate Std. Err. [95% Conf. Interval] Number of poor

Duhouk Qadha Center 299220 0.04 0.02 0.00 0.07 10622

Zawita Nahia 18040 0.03 0.02 0.00 0.08 621

Mankishki Nahia 11669 0.05 0.03 0.00 0.11 607

Sumeil Qadha Center 72531 0.04 0.02 0.00 0.08 2938

Bateel Nahia 22113 0.12 0.04 0.03 0.20 2558

Zakhow Qadha Center 192126 0.06 0.03 0.01 0.11 11566

Rizgary Nahia 21072 0.09 0.04 0.00 0.17 1856

Darkar (AL Syndi) Nahia 16929 0.07 0.03 0.00 0.14 1227

Batifa Nahia 22876 0.07 0.03 0.00 0.13 1533

Al-Amadia Qadha Center 9306 0.03 0.02 0.00 0.07 263

Sarsank Nahia 21785 0.03 0.02 0.00 0.07 741

Kani Massi Nahia 12658 0.05 0.03 0.00 0.12 676

Dirluk Nahia 53465 0.06 0.03 0.00 0.12 3106

Jamanki Nahia 4846 0.04 0.03 0.00 0.10 203

Bamerli Nahia 7459 0.05 0.03 0.00 0.12 374

Qasrouk Nahia 67510 0.10 0.04 0.02 0.18 6683

Atreesh Nahia 12906 0.06 0.04 0.00 0.14 820

Baotheri Nahia 13344 0.09 0.04 0.01 0.17 1182

Akry Qadha Center 64505 0.05 0.02 0.00 0.09 2922

Dinarta Nahia 22972 0.08 0.04 0.00 0.17 1881

Appendix A: Headcount Rate and Number of Poor (by Nahiya)


Appendices

69

TABLE A1: Duhouk

Nahiya Population Headcount rate Std. Err. [95% Conf. Interval] Number of poor

Bijeel Nahia 16681 0.09 0.05 0.00 0.18 1458

Kurdseen Nahia 40772 0.09 0.03 0.03 0.16 3808

Bardarash Qadha Center 26359 0.05 0.03 0.00 0.11 1381

Darto Nahia 30074 0.10 0.04 0.02 0.18 2938

Ro�a Nahia 31480 0.09 0.04 0.02 0.16 2843

kalak Nahia 29623 0.10 0.04 0.01 0.19 2915

(continued)


70

TABLE A2: Nineveh

Nahiya Population Headcount rate Std. Err. [95% Con�dence Interval] Number of poor

Al-Mosal Qadha Center 1289229 0.31 0.07 0.17 0.46 405591

Bashiqa Nahia 110189 0.43 0.07 0.29 0.58 47602

Al-Shoura Nahia 58213 0.48 0.07 0.34 0.62 28082

Hammam Al-Alil 73064 0.42 0.07 0.29 0.56 30884

Al-Qayarra Nahia 127842 0.48 0.06 0.36 0.60 61108

Al-Mahalabia Nahia 34401 0.49 0.07 0.35 0.63 16912

Al-Hamdania Qadha Center 87698 0.20 0.05 0.10 0.29 17162

Namroiud Nahia 62616 0.32 0.07 0.18 0.47 20150

Bartilla Nahia 67056 0.39 0.07 0.25 0.53 26172

Tilkaif Qadha Center 83858 0.27 0.06 0.16 0.39 22935

Wana Nahia 49132 0.26 0.05 0.16 0.37 12976

Alkoush Nahia 63086 0.40 0.06 0.27 0.53 25487

Sinjar Qadha Center 75760 0.37 0.07 0.23 0.51 27986

Al-Shamal Nahia 141128 0.62 0.06 0.49 0.75 87852

Al-Qairawan Nahia 62793 0.58 0.06 0.46 0.70 36527

Telafar Qadha Center 193602 0.49 0.08 0.32 0.66 95078


Appendices

71

TABLE A2: Nineveh


Zummar Nahia 135063 0.50 0.07 0.36 0.64 67788

Rabia Nahia 77796 0.50 0.08 0.35 0.66 39077

Al-Ayadiya Nahia 52393 0.51 0.07 0.37 0.65 26841

Al-Shaikhan Qadha Center 24777 0.13 0.04 0.04 0.22 3209

Zaylakan Nahia 16021 0.28 0.07 0.13 0.43 4467

Al-Hatra Qadha Center 18306 0.31 0.07 0.16 0.45 5594

Al-Tal Nahia 32204 0.40 0.08 0.24 0.56 12856

Al-Baaj Qadha Center 59136 0.64 0.07 0.50 0.77 37817

Al-Qahtanya Nahia 70626 0.60 0.08 0.44 0.76 42708

Makhmoor Qadha Center 46387 0.17 0.05 0.06 0.27 7714

Al-Kuwair Nahia 71330 0.22 0.06 0.11 0.33 15878

Kandinawa Nahia 17584 0.08 0.04 0.00 0.16 1408

Qaraj Nahia 33540 0.33 0.07 0.19 0.46 10911

Mula-Qara Nahia 23818 0.23 0.07 0.10 0.36 5473

Faidah Nahia 75937 0.15 0.05 0.05 0.26 11618

(continued)


72

TABLE A3: Erbil


Arbil Qadha Center 730106 0.03 0.02 0.00 0.07 19786

Bahraka Nahia 48502 0.04 0.02 0.00 0.08 1809

Ainkawa Nahia 25134 0.01 0.01 0.00 0.02 133

Shamamak Nahia 52607 0.07 0.03 0.00 0.13 3593

Dashti Hawler Qadha Center 49736 0.04 0.03 0.00 0.09 1925

Darroo Nahia 52723 0.05 0.04 0.00 0.11 2409

Qoshtaba Nahia 27918 0.07 0.04 0.00 0.15 1854

Kasnasan Nahia 60455 0.05 0.04 0.00 0.13 3156

Sowran Qadha Center 48970 0.04 0.03 0.00 0.10 1915

Khalifan Nahia 33735 0.08 0.04 0.00 0.16 2776

Diyana Nahia 58559 0.05 0.03 0.00 0.10 2998

Siakan Nahia 13094 0.11 0.06 0.00 0.23 1438

Shaqlawa Qadha Center 21205 0.03 0.02 0.00 0.07 568

Salah-eldeen Nahia 47309 0.04 0.02 0.00 0.08 1698

Harir Nahia 32861 0.06 0.03 0.00 0.12 1896

Hyran Nahia 4688 0.05 0.04 0.00 0.12 238

Bamarsa Nahia 17813 0.08 0.05 0.00 0.18 1495

Balisan Nahia 4534 0.08 0.05 0.00 0.19 372

Choman Qadha Center 10457 0.03 0.03 0.00 0.09 365

Haj Omran Nahia 3401 0.06 0.04 0.00 0.13 196

Samilan Nahia 4669 0.07 0.05 0.00 0.16 344

Galala Nahia 1772 0.05 0.04 0.00 0.13 93


Appendices

73

TABLE A3: Erbil


Qasry Nahia 5669 0.05 0.04 0.00 0.12 262

Kuwaisinjaq Qadha Center 56966 0.02 0.02 0.00 0.06 1396

Taq-Taq Nahia 22993 0.06 0.04 0.00 0.14 1423

Shorash Nahia 5947 0.11 0.06 0.00 0.23 635

Ashty Nahia 4199 0.08 0.05 0.00 0.19 354

Saktan Nahia 2371 0.07 0.05 0.00 0.17 164

Sikerdkan Nahia 3401 0.11 0.06 0.00 0.23 367

Meirkasoor Qadha Center 1394 0.04 0.04 0.00 0.11 57

Barazan Nahia 21484 0.05 0.03 0.00 0.11 1100

Biran Nahia 6523 0.06 0.04 0.00 0.13 372

Sherwan Mazin Nahia 3616 0.03 0.03 0.00 0.09 114

Mazni Nahia 5571 0.07 0.04 0.00 0.15 402

Koratoo Nahia 9713 0.03 0.02 0.00 0.08 301

Khabat Qadha Center 38333 0.08 0.05 0.00 0.18 2948

Darashakran Nahia 8446 0.07 0.04 0.00 0.14 609

Rizkari Nahia 35460 0.07 0.04 0.00 0.15 2408

Korakosak Nahia 16358 0.06 0.04 0.00 0.14 1009

Rawandoz Qadha Center 17994 0.05 0.03 0.00 0.10 826

Warny Nahia 5100 0.06 0.04 0.00 0.14 315

(continued)


74

TABLE A4: Sulaimaniya


Sulaimania Qadha Center 598051 0.00 0.00 0.00 0.01 2930

Bakrago Nahia 89179 0.01 0.01 0.00 0.02 847

Bazyan Nahia 36939 0.02 0.02 0.00 0.05 628

Tanjro Nahia 43289 0.02 0.02 0.00 0.06 1043

Qaradagh Qadha Center 7294 0.01 0.01 0.00 0.04 93

Siasitan Nahia 2170 0.01 0.01 0.00 0.04 27

Sharazoor Qadha Center 38712 0.01 0.01 0.00 0.04 538

Warma Nahia 26153 0.02 0.01 0.00 0.04 408

Said Sadiq Qadha Center 72903 0.02 0.02 0.00 0.05 1546

Sarjook Nahia 3676 0.04 0.03 0.00 0.10 161

Halabja Qadha Center 59871 0.00 0.00 0.00 0.01 204

Sirwan Nahia 12329 0.01 0.01 0.00 0.03 129

Khormal Nahia 19787 0.01 0.01 0.00 0.03 186

Biara Nahia 6831 0.00 0.01 0.00 0.02 27

Benjween Qadha Center 25336 0.02 0.01 0.00 0.04 400

Karmak Nahia 10813 0.06 0.04 0.00 0.14 650

Talbraiz Nahia 5989 0.07 0.05 0.00 0.17 418

Jowarta Nahia 8051 0.01 0.01 0.00 0.03 97

Sioutil Nahia 2243 0.01 0.01 0.00 0.04 26

Sitek Nahia 5015 0.02 0.02 0.00 0.05 98

Zalan Nahia 1539 0.02 0.02 0.00 0.06 36

Kabiloon Nahia 3328 0.03 0.03 0.00 0.08 107


Appendices

75



Mout Qadha Center 9230 0.02 0.02 0.00 0.06 186

Qalat-Diza Nahia 67252 0.01 0.01 0.00 0.04 888

Hero Nahia 1833 0.03 0.03 0.00 0.08 47

Halashow Nahia 8488 0.04 0.04 0.00 0.11 334

Tharaow Nahia 12350 0.03 0.02 0.00 0.07 326

Nawadesht Nahia 22015 0.04 0.03 0.00 0.10 896

Eisiway Nahia 2707 0.03 0.03 0.00 0.09 71

Ranya Qadha Center 83175 0.02 0.01 0.00 0.04 1289

Jwar Qorna Nahia 49135 0.02 0.01 0.00 0.04 875

Haji Awa Nahia 55781 0.02 0.02 0.00 0.05 1261

Betwata Nahia 17729 0.03 0.02 0.00 0.08 587

Serkabkan Nahia 7470 0.02 0.02 0.00 0.06 167

Dukan Qadha Center 12308 0.01 0.01 0.00 0.02 94

Sordash Nahia 5291 0.03 0.02 0.00 0.08 156

Bira Magroon Nahia 30431 0.02 0.02 0.00 0.05 630

Khalakan Nahia 4985 0.02 0.01 0.00 0.05 95

Khadran Nahia 2962 0.05 0.03 0.00 0.12 161

Bengrad Nahia 7402 0.05 0.03 0.00 0.10 335

Derbendikhan Qadha Center 42533 0.01 0.01 0.00 0.02 276

Bawkhosheen Nahia 1116 0.01 0.01 0.00 0.04 12

Kalar Qadha Center 131823 0.01 0.01 0.00 0.02 936

Rizgary Nahia 37248 0.02 0.01 0.00 0.04 570

(continued)



76



Bibaz Nahia 7234 0.01 0.01 0.00 0.03 95

Shaikh Taweel Nahia 2520 0.05 0.04 0.00 0.12 128

Kifry Qadha Center 33236 0.01 0.01 0.00 0.04 469

Awaseby Nahia 968 0.15 0.07 0.00 0.29 142

Serqalat Nahia 7979 0.04 0.03 0.00 0.09 299

Neowjool Nahia 1993 0.07 0.04 0.00 0.16 142

Kokas Nahia 7801 0.10 0.04 0.02 0.18 786

Chamchamal Qadha Center 60753 0.01 0.01 0.00 0.03 711

Shorash Nahia 49702 0.02 0.01 0.00 0.05 910

Sangaw Nahia 5680 0.09 0.04 0.00 0.17 490

Takiya Nahia 26767 0.03 0.02 0.00 0.07 822

Aghcheler Nahia 9015 0.05 0.03 0.00 0.12 469

Qader Karam Nahia 2403 0.08 0.04 0.00 0.16 198

Takiya Jabara Nahia 893 0.04 0.04 0.00 0.11 37

Khanaqeen Qadha Center 6654 0.03 0.02 0.00 0.06 172

Bamow Nahia 909 0.03 0.03 0.00 0.10 30

Qoratoo Nahia 6085 0.05 0.03 0.00 0.11 297

(continued)

Appendices

77

TABLE A5: Diyala


Baquba Qadha Center 259528 0.15 0.06 0.01 0.28 37683

Kanan Nahia 42543 0.22 0.08 0.05 0.38 9168

Beni Saad Nahia 130545 0.29 0.09 0.11 0.46 37754

Buhruz (Ashnouna) Nahia 46926 0.17 0.07 0.04 0.30 8071

Al-Abara Nahia 67101 0.20 0.07 0.06 0.35 13642

Al-Muqdadya Qadha Center 148794 0.23 0.08 0.07 0.39 34327

Abi Seda Nahia 38845 0.17 0.07 0.03 0.32 6732

Al-Wajihia Nahia 37260 0.24 0.08 0.06 0.40 8764

Al-Khalis Qadha Center 126554 0.20 0.08 0.02 0.38 25577

Al-Mansuriya Nahia 54557 0.30 0.09 0.11 0.49 16503

Hibhib Nahia 86691 0.18 0.08 0.03 0.34 15977

Al-Sad Al-Adim Nahia 18385 0.29 0.10 0.07 0.49 5240

Al-Salam Nahia 23090 0.22 0.08 0.06 0.37 5057

Khanaqin Qadha Center 82949 0.10 0.05 0.01 0.19 8461

Jalawla Nahia 83675 0.24 0.09 0.05 0.41 19722

Al-Saadiya Nahia 47213 0.24 0.08 0.08 0.40 11454

Baladrooz Qadha Center 89294 0.23 0.08 0.08 0.39 20707

Mendili Nahia 33246 0.32 0.10 0.12 0.52 10788

Qazania Nahia 14541 0.27 0.08 0.10 0.43 3932

Qarataba Nahia 28922 0.17 0.07 0.04 0.30 4943

Jabbara Nahia 7585 0.18 0.07 0.02 0.32 1333


78

TABLE A6: Kirkuk


Kirkuk Qadha Center 886618 0.05 0.01 0.03 0.08 47877

Yaychi Nahia 25903 0.10 0.03 0.04 0.17 2684

Alton Kupry 38224 0.02 0.01 0.00 0.04 849

Al-Multaka (Mula Abdullah)

Nahia

15271 0.19 0.04 0.10 0.27 2866

Taza Khormato Nahia 32208 0.11 0.03 0.05 0.17 3485

Laylan Nahia 19802 0.09 0.03 0.02 0.15 1711

Shwan Nahia 11296 0.05 0.02 0.01 0.10 604

Qara Hanjeer (Al-Rabee)

Nahia

10949 0.05 0.02 0.01 0.10 558

Al-Hawiga Qadha Center 95854 0.13 0.03 0.07 0.18 12020

Al-Abbasi Nahia 49339 0.23 0.05 0.14 0.33 11560

Al-Riyadh Nahia 44931 0.25 0.05 0.15 0.34 11022

Al-Zab Nahia 52758 0.16 0.04 0.09 0.23 8383

Daquq Qadha Center 54860 0.15 0.03 0.08 0.21 8229

Al-Rashad Nahia 23204 0.27 0.05 0.17 0.36 6163

Dibis Qadha Center 45048 0.13 0.03 0.07 0.19 5888

Sarkran (Al-Qudis) Nahia 18037 0.10 0.03 0.04 0.15 1719

Appendices

79

TABLE A7: Anbar


Al-Ramadi Qadha Center 368374 0.15 0.05 0.05 0.25 54556

Al-Habbaniya Nahia 136466 0.16 0.05 0.06 0.26 21589

Al-wa�a Nahia 9749 0.48 0.10 0.27 0.69 4702

Heet Qadha Center 80563 0.17 0.05 0.07 0.26 13663

Baghdady Nahia 28616 0.12 0.05 0.02 0.23 3497

Kubaisa Nahia 14632 0.14 0.05 0.05 0.23 2035

Al-Forat Nahia 24713 0.28 0.07 0.13 0.42 6937

Al-Falluja Qadha Center 293877 0.17 0.05 0.08 0.26 49107

Al-Amirya Nahia 78756 0.13 0.05 0.04 0.23 10412

Al-Saklawiya Nahia 49361 0.19 0.07 0.06 0.32 9522

Al-Garma Nahia 139664 0.46 0.12 0.21 0.70 63757

Ana Qadha Center 31336 0.08 0.03 0.02 0.14 2479

Haditha Qadha Center 50095 0.11 0.04 0.03 0.19 5606

Al-Haqlaniya Nahia 29185 0.09 0.03 0.03 0.15 2542

Barwana Nahia 27838 0.13 0.04 0.05 0.20 3491

Al-Rutba Qadha Center 32035 0.25 0.06 0.13 0.35 7858

Al-Walid Nahia 6221 0.09 0.04 0.01 0.18 577

Al-Nakhaeb Nahia 2975 0.18 0.05 0.07 0.29 539

Al-Kaim Qadha Center 95636 0.22 0.05 0.11 0.32 20830

Al-Obour Nahia 36089 0.28 0.08 0.13 0.44 10260

Al-Obiadi Nahia 29511 0.08 0.04 0.01 0.16 2500

Rawa Qadha Center 23083 0.06 0.02 0.01 0.10 1304


80

TABLE A8: Salahadin


Tikrit Qadha Center 167383 0.15 0.05 15.68 0.15 24940

Alam Nahia 52988 0.15 0.05 16.16 0.15 7911

Tooz-Khormato Qadha

Center

108301 0.13 0.05 14.47 0.13 13538

Amerly Nahia 42913 0.20 0.07 13.92 0.20 8793

Sulaiman Baig Nahia 24751 0.21 0.08 14.35 0.21 5091

Samarra Qadha Center 182139 0.14 0.05 16.43 0.14 25427

Al-Muotasim Nahia 16438 0.18 0.07 14.20 0.18 2919

Dijla Nahia 16076 0.20 0.07 14.03 0.20 3199

Balad Qadha Center 73122 0.14 0.05 16.75 0.14 10237

Al-Dholoia Nahia 57863 0.18 0.06 14.09 0.18 10583

Al-Eshaki Nahia 41193 0.21 0.06 12.08 0.21 8457

Yathrib Nahia 73627 0.20 0.06 13.36 0.20 14512

Beygee Qadha Center 170961 0.17 0.06 15.78 0.17 29354

Al-Ssynia Nahia 35680 0.17 0.06 16.70 0.17 5962

Al-Daur Qadha Center 56181 0.09 0.04 19.36 0.09 5112

Al-Shirqat Qadha Center 199567 0.18 0.06 15.29 0.18 35164

Al-Dijail Qadha Center 106129 0.20 0.05 14.29 0.20 21194

Appendices

81

TABLE A9: Baghdad


Al-Rusafa Qadha Center 114797 0.04 0.01 0.01 0.06 4328

Al-Karrada Al-Sharqia Nahia 311485 0.03 0.01 0.01 0.04 7818

Baghdad Al-Jedeeda Nahia 1140276 0.11 0.02 0.07 0.15 126000

Palestine Nahia 84777 0.00 0.00 0.00 0.01 322

Al-Adhamia Qadha Center 299280 0.02 0.01 0.01 0.03 6464

Al-Rashdia Nahia 41847 0.16 0.02 0.11 0.21 6733

Al-Fahama Nahia 614606 0.12 0.02 0.08 0.16 73568

Al-Zohour Nahia 198287 0.24 0.04 0.17 0.31 47728

Sader /2 Qadha Center 39683 0.08 0.02 0.04 0.12 3060

Abna’a Al-Ra�dain Nahia 140824 0.13 0.02 0.08 0.17 17913

Al-Mounawara Nahia 262843 0.18 0.03 0.13 0.23 48100

Sader /1 Qadha Center 107069 0.11 0.02 0.07 0.15 11446

Al-Sideeq Al-Akbar Nahia 139347 0.12 0.02 0.07 0.16 16039

Al-Forat Nahia 295399 0.08 0.02 0.05 0.11 23868

Al-Karkh Qadha Center 99483 0.02 0.01 0.01 0.03 2020

Al-Mansour Nahia 441714 0.02 0.01 0.01 0.04 10071

Al-Mamoon Nahia 856502 0.03 0.01 0.01 0.05 26295

Al-Kadimiya Qadha Center 380306 0.05 0.01 0.02 0.07 18103

That Al Salasil Nahia 249432 0.13 0.02 0.09 0.17 31753

Al-Taji Nahia 157441 0.22 0.03 0.16 0.27 34149

Mahmudiya Qadha Center 132673 0.11 0.02 0.07 0.14 14024

Al-Yousifya Nahia 128003 0.30 0.03 0.23 0.36 37914



82

TABLE A9: Baghdad


Al-Latifya Nahia 90228 0.18 0.03 0.13 0.23 16142

Al-Rasheed Nahia 70755 0.18 0.03 0.13 0.23 12863

Abu-Gharib Qadha Center 149233 0.17 0.03 0.12 0.22 24758

Al-Nasir & Al-Salam Nahia 150969 0.27 0.04 0.20 0.35 41335

Al-Tarmiya Qadha Center 74957 0.19 0.03 0.14 0.25 14444

Al-Mishahda Nahia 24696 0.49 0.05 0.39 0.60 12212

Al-Abiaji Nahia 13542 0.11 0.02 0.06 0.15 1430

Al-Mada’in Qadha Center 60739 0.07 0.02 0.03 0.11 4300

Al-Jisr Nahia 154422 0.18 0.03 0.12 0.24 27286

Al-Wihda Nahia 217970 0.27 0.04 0.19 0.35 58634

(continued)

Appendices

83

TABLE A10: Babylon


Al-Hilla Qadha Center 540313 0.06 0.02 0.02 0.10 34688

Al-Ki� Nahia 143532 0.20 0.05 0.10 0.29 28333

Abi Gharaq Nahia 107182 0.12 0.03 0.05 0.19 12551

Al-Mahawil Qadha Center 97792 0.09 0.03 0.03 0.15 9124

Al-Mashroo Nahia 107289 0.19 0.04 0.11 0.27 20771

Al-Emam Nahia 31725 0.11 0.04 0.04 0.19 3604

Nile Nahia 55976 0.19 0.05 0.09 0.29 10742

Al-Hashimiya Qadha Center 34816 0.08 0.03 0.01 0.15 2803

Al-Qasim Nahia 154236 0.16 0.04 0.07 0.25 24878

Al-Madhatiya Nahia 123158 0.16 0.04 0.07 0.25 19989

Al-Shomaly Nahia 77233 0.21 0.05 0.11 0.30 15972

Al-Talea’a Nahia 39739 0.24 0.06 0.11 0.35 9430

Al-Mussyab Qadha Center 51656 0.07 0.03 0.01 0.12 3497

Saddat Al-Hindin Nahia 98499 0.10 0.03 0.04 0.17 10333

Jurf Al-Sakhar Nahia 40310 0.11 0.04 0.03 0.18 4249

Al-Iskandaria Nahia 148019 0.10 0.03 0.03 0.17 14624


84

TABLE A11: Kerbala


Kerbela Qadha Center 478879 0.12 0.03 0.07 0.17 55789

Al-Hassainya Nahia 138949 0.09 0.02 0.04 0.13 12005

Al-Hur Nahia 211975 0.20 0.04 0.13 0.27 41929

Ain Al-Tamur Qadha Center 26924 0.13 0.04 0.06 0.21 3592

Al-Hindiya Qadha Center 106961 0.10 0.02 0.05 0.15 10557

Al-Jadwal Al-Ghrabi Nahia 75029 0.18 0.04 0.11 0.25 13565

Al-Kharirat Nahia 50715 0.16 0.04 0.09 0.23 8099

Appendices

85

TABLE A12: Wasit


Al-Kut Qadha Center 407511 0.26 0.06 0.15 0.38 107746

Wasit Nahia 39912 0.43 0.08 0.27 0.58 17050

Shaekh Saad Nahia 37237 0.48 0.08 0.33 0.63 17855

Al-Noamaniya Qadha Center 99503 0.18 0.05 0.08 0.28 17682

Al-Ahrar Nahia 45482 0.23 0.06 0.11 0.36 10661

Al-Hai Qadha Center 86108 0.14 0.03 0.07 0.21 12313

Al-Mowafaqiya Nahia 45903 0.32 0.05 0.22 0.43 14836

Al-Bashaer Nahia 36733 0.35 0.06 0.23 0.47 12930

Badra Qadha Center 17372 0.27 0.06 0.14 0.39 4635

Jassan Nahia 10741 0.33 0.07 0.19 0.48 3583

Al-Tahab Nahia (Zerbattiya) 684 0.22 0.09 0.03 0.40 148

Al-Suwaira Qadha Center 136628 0.14 0.04 0.07 0.22 19592

Al-Zubaidiya Nahia 53975 0.16 0.04 0.08 0.25 8803

Al-Shehamiya Nahia 34095 0.22 0.06 0.11 0.33 7532

Al-Aziziya Qadha Center 93164 0.24 0.06 0.12 0.35 22071

Taj-eldeen (AlHafriya) Nahia 75204 0.28 0.07 0.15 0.41 21065

Al-Duboni Nahia 18118 0.37 0.08 0.22 0.52 6747


86

TABLE A13: Najaf


Al-Najaf Qadha Center 688448 0.06 0.02 0.03 0.09 43166

Al-Haydariya Nahia 45096 0.19 0.03 0.12 0.26 8699

Al-Shabaka Nahia 356 0.16 0.07 0.01 0.31 57

Al-Kufa Qadha Center 217736 0.07 0.02 0.04 0.11 16200

Al-Abbassiya Nahia 84544 0.10 0.03 0.04 0.15 8125

Al-Huriya Nahia 29087 0.13 0.03 0.07 0.19 3912

Al-Manathera Qadha Center 82451 0.13 0.03 0.08 0.18 10677

Al-Heera Nahia 35223 0.12 0.03 0.06 0.18 4114

Al-Mishkhab Nahia 83299 0.12 0.03 0.07 0.17 9879

Al-Qadisiya Nahia 47066 0.21 0.04 0.12 0.28 9766

Appendices

87

TABLE A14: Qadisiya


Al-Diwaniya Qadha Center 388415 0.27 0.04 0.20 0.35 105998

Al-Saniya Nahia 44312 0.36 0.05 0.25 0.46 15775

Al-Shafeia Nahia 44034 0.35 0.05 0.25 0.44 15227

Al-Daghara Nahia 58930 0.44 0.06 0.33 0.55 26023

Afaq Qadha Center 52169 0.56 0.05 0.45 0.66 29136

Nafar Nahia 19527 0.44 0.06 0.32 0.56 8584

Al-Badair Nahia 60219 0.65 0.05 0.54 0.75 38974

Sumer Nahia 32587 0.34 0.06 0.22 0.45 10936

Al-Shamiya Qadha Center 94112 0.51 0.05 0.42 0.60 48176

Gammas Nahia 87526 0.77 0.04 0.68 0.83 67351

Al-Mihanawiya Nahia 42255 0.45 0.06 0.34 0.57 19222

Al-Salahiya Nahia 31130 0.61 0.06 0.48 0.73 18834

Al-Hamza Qadha Center 112885 0.56 0.05 0.46 0.67 63385

Al-Sadeer Nahia 40584 0.57 0.05 0.47 0.67 23222

Al-Shina�ya Nahia 49401 0.57 0.06 0.45 0.68 28109


88

TABLE A15: Muthanna


Al-Samawa Qadha Center 271285 0.41 0.10 0.19 0.62 111254

Al-Sowair Nahia 39066 0.70 0.09 0.51 0.87 27428

Al-Rumaitha Qadha Center 118423 0.49 0.12 0.23 0.74 57909

Al-Majd Nahia 37727 0.66 0.09 0.46 0.84 24855

Al-Warka Nahia 93216 0.71 0.09 0.52 0.88 65988

Al-Najmi Nahia 32461 0.72 0.08 0.54 0.88 23453

Al-Hilal Nahia 31800 0.73 0.08 0.55 0.88 23255

Al-Salman Qadha Center 7818 0.58 0.12 0.34 0.80 4506

Al-Bussaiya Nahia 927 0.57 0.12 0.32 0.80 527

Al-Khdhir Qadha Center 77965 0.58 0.11 0.35 0.79 44908

Al-Daraji Nahia 18258 0.70 0.09 0.51 0.86 12740

Appendices

89

TABLE A16: Thi Qar


Al-Nasiriya Qadha Center 498661 0.23 0.04 0.14 0.32 115141

Al-Islah Nahia 40695 0.37 0.06 0.25 0.49 15110

Al-Battha’a Nahia 40259 0.35 0.06 0.23 0.48 14252

Said Dekhil Nahia 55881 0.63 0.05 0.53 0.72 35172

Ur Nahia 58637 0.41 0.07 0.27 0.54 23848

Al-Rifaai Qadha Center 130791 0.38 0.06 0.27 0.49 49596

Qalat Siker Nahia 100416 0.46 0.06 0.34 0.58 46623

Al-Nasr Nahia 84910 0.37 0.06 0.25 0.48 31289

Al-Fajer Nahia 55939 0.46 0.06 0.34 0.58 25877

Suq AL-Shoyolh Qadha

Center

117837 0.32 0.06 0.21 0.43 37684

Akaika Nahia 45459 0.49 0.07 0.35 0.62 22061

Garmat Beni Said Nahia 57436 0.42 0.07 0.29 0.55 24066

Al-Fadhliya Nahia 53086 0.33 0.06 0.20 0.45 17370

Al-Ttar Nahia 18161 0.49 0.08 0.33 0.64 8830

Al-Chibayish Qadha Center 41831 0.53 0.06 0.41 0.64 22175

Al-Hammer Nahia 8521 0.55 0.06 0.43 0.67 4726

Al-Fhood Nahia 44075 0.48 0.06 0.35 0.61 21182

Al-Shattra Qadha Center 227380 0.37 0.06 0.25 0.49 84631

Al-Dawaya Nahia 77893 0.62 0.05 0.51 0.71 48216

Al-Gharraf Nahia 111957 0.48 0.07 0.34 0.62 54064


90

TABLE A17: Missan

Qhada Population Headcount rate Std. Err. [95% Con�dence Interval] Number of poor

Al-Amara Qadha Center 529251 0.25 0.08 0.09 0.41 131625

Kumait Nahia 26709 0.44 0.10 0.23 0.64 11755

Ali Al-Garbi Qadha Center 31250 0.30 0.10 0.11 0.50 9481

Ali Al-Sharqi Nahia 17859 0.21 0.09 0.02 0.40 3727

Al-Maimouna Qadha Center 52379 0.48 0.07 0.33 0.62 24896

Al-Salam Nahia 35140 0.66 0.06 0.54 0.76 23217

Said Ahmed Al-Rifaai Nahia 10450 0.67 0.08 0.50 0.82 6968

Qalat Saleh Qadha Center 53955 0.68 0.06 0.55 0.77 36808

Al-Ezair Nahia 29933 0.70 0.05 0.58 0.79 20875

Al-Mejar Al-Kabir Qadha

Center

84476 0.38 0.07 0.24 0.52 32067

Al-Adel Nahia 15243 0.25 0.08 0.10 0.40 3838

Al-Khayr Nahia 19585 0.64 0.07 0.49 0.76 12446

Al-Kahla’a Qadha Center 37435 0.44 0.07 0.30 0.58 16412

Al-Mshsrsh Nahia 30369 0.47 0.07 0.33 0.60 14176

Bani Hashim Nahia 20003 0.72 0.06 0.60 0.83 14498

Appendices

91

TABLE A18: Basrah


Al-Basrah Qadha Center 1172742 0.10 0.02 0.06 0.13 111880

Al-Hartha Nahia 140123 0.12 0.02 0.07 0.16 16633

Abu Al-Khaseeb Qadha

Center

190855 0.14 0.02 0.09 0.18 25842

Al-Zubair Qadha Center 353932 0.18 0.02 0.14 0.23 65088

Safwan Nahia 36065 0.18 0.03 0.12 0.24 6470

Um Qasr Nahia 51314 0.05 0.02 0.02 0.09 2776

Al-Qurna Qadha Center 119267 0.15 0.03 0.10 0.20 18331

Al-Dair Nahia 90761 0.26 0.03 0.19 0.32 23416

Al-Thagar Nahia 35542 0.28 0.05 0.19 0.37 10126

Al-Faw Qadha Center 34766 0.09 0.02 0.04 0.13 2979

Shat Al-Arab Qadha Center 123049 0.11 0.02 0.07 0.16 14089

Al-Nashwa Nahia 29167 0.25 0.04 0.17 0.33 7265

Al-Midaina Qadha Center 68677 0.15 0.03 0.10 0.20 10192

Iz-Eldeen Salim Nahia 60926 0.17 0.03 0.10 0.23 10236

Talha (Al-Sadiq) Nahia 66836 0.11 0.03 0.06 0.17 7673


92

Appendix B: GLS Models

Duhok Coe�cient Std. Err. t |Prob|>t

Intercept 5.4114 0.0402 134.6406 0

Dummy variable for ownership of air conditioner 0.1433 0.0247 5.8052 0

Dummy variable for age 30–39 0.0624 0.0263 2.3736 0.0178

Dummy variable for age 50–59 –0.0784 0.03 –2.6111 0.0091

Number of children between ages 7 and 17 –0.1061 0.0087 –12.1801 0

Dependency ratio –0.2417 0.0892 –2.7079 0.0069

Dummy variable for primary level complete 0.0599 0.0267 2.2462 0.0249

Dummy variable for secondary level incomplete or complete 0.1985 0.0761 2.6095 0.0092

Dummy variable for institute level complete 0.151 0.0479 3.1534 0.0017

Dummy variable for graduate/technical level or post graduate level complete 0.2389 0.0479 4.992 0

Number of household members aged 60 or above –0.1614 0.0286 –5.6445 0

Proportion of household members between ages 0 and 6 –0.5535 0.1056 –5.2395 0

Proportion of household members aged 60 or above 0.5871 0.1612 3.6428 0.0003

Dummy variable for ownership of vacuum cleaner 0.0884 0.0236 3.7547 0.0002

Appendices

93

Nainawa Coe�cient Std. Err. t |Prob|>t

Intercept 4.481 0.1137 39.4164 0

Dummy variable for age 20–29 0.1584 0.0364 4.3477 0

Car ownership rate in nahiya 0.9682 0.2374 4.0785 0

Dependency ratio –0.6164 0.05 –12.3298 0

Proportion of household heads with secondary level incomplete or complete in

nahiya

–2.8675 0.827 –3.4673 0.0005

Dummy variable for institute level complete 0.3067 0.0433 7.0791 0

Dummy variable for graduate/technical level or post graduate level complete 0.4704 0.0402 11.7007 0

Dummy variable for male –0.1125 0.0372 –3.0269 0.0025

Dummy variable for maximum education in the household is intermediate level

incomplete or complete

–0.0703 0.0241 –2.9107 0.0037

Dummy variable for employment in government/public sector 0.0945 0.0233 4.0542 0.0001

Proportion of household members aged 60 or above 0.4306 0.1053 4.0877 0

Share of working age males employed 0.1632 0.0282 5.7844 0

Proportion of households with vacuum in nahiya 0.7021 0.2194 3.1999 0.0014

Dummy variable for brick wall 0.202 0.0651 3.1056 0.0019

Interaction term for rural area and dummy variable for age of household head

40–49

–0.1254 0.0307 –4.0851 0

Interaction term for urban area and dummy variable for age of household head 70

or more

0.1986 0.062 3.2043 0.0014

Interaction term for urban area and dummy variable for intermediate level


–0.1578 0.034 –4.6363 0

Interaction term for urban area and dummy variable for secondary level


0.223 0.051 4.3697 0

Interaction term for urban area and dummy variable for maximum education in

the household is institute level complete

0.1128 0.0368 3.0635 0.0022



94

Nainawa Coe�cient Std. Err. t |Prob|>t

Interaction term for rural area and dummy variable for maximum education in the

household is graduate/technical level or post graduate level complete

–0.3075 0.0546 –5.6369 0

Interaction term for urban area and dummy variable for principal material of wall is

cement block, concrete readymade, or precast

–0.175 0.0303 –5.7833 0

Interaction term for rural area and main source of drinking water is public network –0.1054 0.0345 –3.0542 0.0023

(continued)

Appendices

95

Erbil Coe�cient Std. Err. t |Prob|>t

Intercept 5.2698 0.0528 99.761 0


Dummy variable for household head’s age 60–69 –0.1976 0.0347 –5.6931 0

Dummy variable for household head’s age 70 or more –0.1776 0.0485 –3.6658 0.0003



Access of electricity from private generator (days per week) 0.0226 0.0081 2.7962 0.0052

Dummy variable for household head has no education –0.0886 0.0208 –4.2684 0

Dummy variable for household head’s education is institute level complete 0.1372 0.0329 4.1712 0

Dummy variable for household head’s education is graduate/technical level or

post graduate level complete

0.1691 0.035 4.8348 0

Number of males employed –0.0796 0.0258 –3.0803 0.0021

Dummy variable for ownership of freezer 0.0512 0.019 2.6969 0.0071

Dummy variable for dwelling owned by household 0.0779 0.0207 3.7683 0.0002

Dummy variable for male –0.1134 0.0275 –4.1189 0


Dummy variable for principal material of wall is brick 0.1217 0.0308 3.9504 0.0001

Dummy variable for ownership of washing machine 0.1047 0.0301 3.482 0.0005


96

Sulaimaniya Coe�cient Std. Err. t |Prob|>t

Intercept 6.1924 0.0524 118.0758 0


Dummy variable for household head’s age 40–49 0.065 0.0146 4.4488 0

Dummy variable for household head’s age 70 or more –0.1832 0.0193 –9.5032 0

Dummy variable for household head has no education –0.077 0.0136 –5.6396 0

Number of males employed = 0 –0.1912 0.0322 –5.9331 0

Number of males employed = 2 0.0646 0.0188 3.4324 0.0006

Number of males employed = 3 0.104 0.0312 3.3331 0.0009

Dummy variable for ownership of freezer 0.1654 0.0133 12.4547 0

Household size –0.276 0.0107 –25.8669 0

Square of the household size 0.0132 0.0007 17.6451 0

Dummy variable for maximum education in the household is primary level

complete

0.0957 0.0212 4.5099 0



0.1586 0.0237 6.6977 0

Dummy variable for maximum education in the household is secondary level


0.1035 0.0234 4.4222 0

Dummy variable for maximum education in the household is institute level

complete

0.136 0.026 5.2415 0

Dummy variable for maximum education in the household is graduate/technical

level or post graduate level complete

0.1841 0.0252 7.3185 0

Dummy variable for source of electricity is private generator 0.1207 0.0194 6.225 0



Appendices

97

Sulaimaniya Coe�cient Std. Err. t |Prob|>t


Share of working age males employed –0.1511 0.0318 –4.7537 0

Dummy variable for ownership of vacuum cleaner 0.1446 0.0193 7.4855 0

Dummy variable for principal material of wall is cement block, concrete

readymade, or precast

0.0478 0.0155 3.0842 0.0021

(continued)


98

Diyala Coe�cient Std. Err. t |Prob|>t

Intercept 5.5157 0.0646 85.3229 0


Number of children between ages 0 and 6 –0.0325 0.0108 –3.017 0.0026

Access of electricity from shared generator is �ve days per week 0.392 0.131 2.9914 0.0028

Dependency ratio –0.1926 0.0539 –3.5721 0.0004

Dummy variable for household head’s education is primary level incomplete 0.093 0.0338 2.7564 0.0059


Dummy variable for ownership of fridge 0.0859 0.0381 2.2573 0.0242

Dummy variable for ownership of generator 0.1577 0.0204 7.715 0

Household size –0.2123 0.0126 –16.8536 0

Square of household size 0.0089 0.0008 11.2718 0



0.0744 0.0274 2.715 0.0067



Appendices

99

Kirkuk Coe�cient Std. Err. t |Prob|>t

Intercept 4.5162 0.0534 84.5791 0


Dummy variable for household head’s age 20–29 0.1218 0.049 2.4832 0.0132


Number of children between ages 7 and 17 = 0 0.4608 0.0409 11.2771 0


Number of children between ages 7 and 17 = 5 –0.1679 0.0735 –2.2846 0.0226

Number of children between ages 7 and 17 = 6 –0.2162 0.0969 –2.2308 0.026

Dependency ratio 0.3195 0.0912 3.504 0.0005


Dummy variable for dwelling owned by government/public sector –0.2496 0.0367 –6.7916 0

Dummy variable for male 0.2447 0.0364 6.7207 0



–0.0429 0.0294 –1.46 0.1447


complete

0.1868 0.0401 4.6583 0

Dummy variable for employment in private sector –0.0835 0.0231 –3.6068 0.0003




Dummy variable for principal material of wall is clay, bamboo, or other –0.2311 0.0332 –6.9655 0

Dummy variable for ownership of washing machine 0.1607 0.0314 5.1105 0


100

Anbar Coe�cient Std. Err. t |Prob|>t

Intercept 9.5194 1.5028 6.3343 0



Dummy variable for household head’s age 70 or more 0.2148 0.0316 6.7944 0

Access of electricity from private generator (days per week) 0.0207 0.0027 7.7264 0

Dummy variable for household head’s education is primary level incomplete –0.0522 0.0172 –3.0391 0.0024



0.0749 0.0263 2.8447 0.0045


Dummy variable for maximum education in the household is secondary level


–0.0593 0.015 –3.9463 0.0001

Proportion of household heads employed in the private sector in nahiya –0.5733 0.1907 –3.0064 0.0027


Proportion of households with television in nahiya –4.9148 1.508 –3.2592 0.0011



Interaction of rural area and proportion of household members between ages 0

and 6

–0.1414 0.0585 –2.4175 0.0157

Interaction of urban area and proportion of household members between ages 0

and 6

–0.4129 0.0632 –6.532 0

Interaction of rural area and dummy variable for household head’s age 20–29 0.2208 0.048 4.5993 0

Interaction of rural area and dummy variable for ownership of cooler –0.2111 0.0695 –3.037 0.0024


Appendices

101

Anbar Coe�cient Std. Err. t |Prob|>t

Interaction of urban area and dummy variable for household head has no

education

–0.1797 0.0298 –6.0261 0

Interaction of urban area and dummy variable for household head’s education is

primary level complete

0.0535 0.0202 2.6449 0.0082

Interaction for rural area and no ownership of freezer –0.1112 0.0199 –5.5849 0

Interaction of rural area and household head is unmarried –0.1788 0.0388 –4.6122 0

Interaction of rural area and dummy variable for maximum education in the

household is no education

–0.4597 0.0913 –5.0346 0

Interaction of rural area and no ownership of water heater 0.1421 0.041 3.4651 0.0005

(continued)


102

Salahaddin Coe�cient Std. Err. t |Prob|>t

Intercept 4.7414 0.0576 82.3636 0

Dummy variable for ownership of car 0.1498 0.0211 7.108 0


Dummy variable for household head’s education is intermediate level incomplete

or complete

0.0704 0.0313 2.2487 0.0247

Dummy variable for household head’s education is institute level complete 0.0786 0.0344 2.2858 0.0224



0.2015 0.0307 6.571 0



Dummy variable for ownership of generator 0.0595 0.0284 2.0952 0.0363

Dummy variable for married –0.0973 0.0362 –2.6862 0.0073

Dummy variable for source of electricity is private generator 0.1026 0.0294 3.4835 0.0005

Dummy variable for public sewage network 0.0773 0.0286 2.7 0.007

Proportion of household members between ages 7 and 17 0.593 0.091 6.517 0

Dummy variable for source of electricity is shared generator 0.1428 0.0311 4.5974 0



Dummy variable for principal material of wall is clay, bamboo, or other –0.1606 0.0332 –4.8437 0

Dummy variable for main source of drinking water is public network –0.1386 0.0361 –3.8377 0.0001

Dummy variable for ownership of water heater 0.0911 0.0344 2.6491 0.0081

Appendices

103

Baghdad Coe�cient Std. Err. t |Prob|>t

Intercept 5.1627 0.1438 35.8925 0


Dummy variable for household head’s age 40–49 –0.0806 0.0202 –3.9872 0.0001

Access of electricity from shared generator (days per week) 0.0209 0.0053 3.9125 0.0001

Dummy variable for household head has no education –0.0878 0.0271 –3.2369 0.0012

Proportion of household heads with primary level incomplete in nahiya –0.5831 0.1813 –3.2153 0.0013

Dummy variable for household head’s education is secondary level incomplete or

complete

0.1531 0.0285 5.3672 0




0.1943 0.0357 5.4473 0


Proportion of households with generator in nahiya –0.2233 0.0723 –3.0866 0.0021

Average household size in nahiya –0.0786 0.0194 –4.0643 0

Dummy variable for maximum education in the household is no education 0.132 0.0596 2.215 0.0269


complete

0.0861 0.0355 2.4289 0.0152



0.115 0.028 4.0986 0

Average distance to the nearest road in nahiya –0.0228 0.0041 –5.6366 0

Average distance to the nearest school in nahiya 0.0146 0.0047 3.126 0.0018

Average distance to the nearest school in qhada 0.0206 0.0068 3.0515 0.0023

Dummy variable for household head is unemployed –0.0816 0.0213 –3.8406 0.0001


Dummy variable for closed drain, open drain, or other –0.099 0.0426 –2.3211 0.0204



104

Baghdad Coe�cient Std. Err. t |Prob|>t





–0.0945 0.0294 –3.2141 0.0013

(continued)

Appendices

105

Babylon Coe�cient Std. Err. t |Prob|>t

Intercept 4.9265 0.0485 101.5122 0










0.1962 0.0459 4.2711 0

Dummy variable for housing unit is clay house, bamboo house, or other –0.2327 0.0997 –2.3331 0.0199




–0.0761 0.0346 –2.1962 0.0284


Proportion of household members between ages 7 and 17 0.2961 0.1133 2.6143 0.0091





106

Kerbala Coe�cient Std. Err. t |Prob|>t

Intercept 5.6676 0.0662 85.5919 0



DAYS_SHARED_GENERATOR_6 Days 0.4051 0.1575 2.5721 0.0104



0.0717 0.0591 1.2141 0.2252

Number of employed male = 3 0.1136 0.0289 3.9327 0.0001

Number of employed = 5 –0.2476 0.0376 –6.5855 0


Household size –0.1925 0.0082 –23.3487 0




0.2979 0.0346 8.6188 0

Dummy variable for public sewage network 0.1022 0.0217 4.7115 0

Proportion of household members between ages 7 and 17 –0.1373 0.0486 –2.8237 0.0049


Dummy variable for principal material of wall is brick 0.1056 0.0176 5.9929 0


Dummy variable for main source of drinking water is public network –0.1681 0.0372 –4.5153 0

Interaction of urban area and proportion of household members between ages 0

and 6

–0.3987 0.0654 –6.0936 0

Interaction of urban area and dummy variable for born outside the governorate 0.0838 0.0222 3.7738 0.0002

Interaction of rural area and dummy variable for household head has no education 0.1982 0.0287 6.8934 0


Appendices

107

Kerbala Coe�cient Std. Err. t |Prob|>t

Interaction of urban area and dummy variable for household head’s education is

primary level incomplete

–0.118 0.0277 –4.2607 0

Interaction of rural area and dummy variable for household head’s education is

graduate/technical level or post graduate level complete

–0.4603 0.0828 –5.5616 0

Interaction for rural area and number of employed males = 2 0.1744 0.032 5.4438 0

Interaction of rural area and no fridge ownership –0.0994 0.0313 –3.1785 0.0016

Interaction of urban area and dummy variable for dwelling owned by

government/public sector

–0.2024 0.037 –5.4674 0

Interaction for rural area and dummy variable for not married –0.1223 0.044 –2.7778 0.0057

Interaction for urban area and dummy variable for maximum education in the

household is institute level complete

0.2218 0.0381 5.8234 0

Interaction for urban area and dummy variable for maximum education in the

household is graduate/technical level or post graduate level complete

–0.2663 0.042 –6.3386 0

Interaction of rural area and dummy variable for no public sewage network –0.0525 0.0262 –2.0015 0.0458

Interaction of urban area and dummy variable for no public sewage network –0.1196 0.0848 –1.4103 0.159

(continued)


108

Wasit Coe�cient Std. Err. t |Prob|>t

Intercept 4.5919 0.1104 41.5748 0





Dependency ratio 0.229 0.1087 2.1058 0.0354

Dummy variable for household head’s education is secondary level incomplete or

complete

0.1687 0.0473 3.5675 0.0004

Dummy variable for household head’s education is institute level complete 0.2637 0.0426 6.1918 0



0.3942 0.0573 6.8851 0



–0.1478 0.0409 –3.612 0.0003

Dummy variable for house provided by employer, free with and without

arrangement with owner, random housing

–0.0566 0.0326 –1.7354 0.0829

Dummy variable for source of electricity is private generator 0.1602 0.0262 6.1167 0

Average distance to the nearest road in qhada 0.0502 0.0148 3.3957 0.0007

Dummy variable for household head is employed in the government/public sector 0.1399 0.0293 4.7745 0




Dummy variable for principal material of wall is clay, bamboo, or other –0.1518 0.0403 –3.7643 0.0002


Appendices

109

Najaf Coe�cient Std. Err. t |Prob|>t

Intercept 4.9407 0.0898 55.0178 0


Dummy variable for household head’s age 70 or more –0.3048 0.0625 –4.8803 0


Access of electricity from shared generator (days per week) 0.0227 0.0085 2.676 0.0076

Dummy variable for household head’s education is primary level incomplete –0.0867 0.042 –2.0647 0.0394

Dummy variable for closed drain, open drain, or other –0.1819 0.0496 –3.6683 0.0003






Dummy variable for ownership of water heater 0.1324 0.0395 3.3496 0.0009


110

Qadisiyah Coe�cient Std. Err. t |Prob|>t

Intercept 4.1158 1.091 3.7725 0.0002



Dummy variable for household head’s age 70 or more 0.1722 0.0561 3.0725 0.0022

Dummy variable for born outside the governorate –0.1563 0.0426 –3.6696 0.0003


Proportion of households owning cooker in nahiya –0.7713 0.1731 –4.4556 0



Number of males employed = 1 0.1017 0.0239 4.2615 0

Number of males employed = 3 –0.0803 0.0452 –1.7754 0.0762


Dummy variable for housing unit is house or �at 0.1253 0.0426 2.9432 0.0033

Dummy variable for ownership of personal computer 0.2552 0.0411 6.2037 0


Proportion of household heads employed in government/public sector in nahiya 1.5853 0.3441 4.6066 0

Dummy variable for septic tank 0.1357 0.0253 5.3648 0



Proportion of households with television in nahiya 0.3725 1.1976 0.311 0.7559




Dummy variable for main source of drinking water is river/canal/creek/wheel, open

well/covered well, pond/lake, spring, kehriz (man built spring) or other

0.2107 0.0387 5.4397 0

Appendices

111

Muthana Coe�cient Std. Err. t |Prob|>t

Intercept 4.5534 0.0606 75.1531 0


Dummy variable for ownership of cooler –0.0304 0.0272 –1.1182 0.2638





0.2176 0.0669 3.2553 0.0012



complete

0.0343 0.0295 1.1609 0.246

Dummy variable for household head is employed in the government/public sector 0.0723 0.031 2.3328 0.0199

Dummy variable for closed drain, open drain, or other –0.1159 0.0257 –4.5101 0


Proportion of household members aged 60 or above 0.6669 0.2058 3.2408 0.0012




–0.0802 0.0247 –3.2477 0.0012



112

Thi Qar Coe�cient Std. Err. t |Prob|>t

Intercept 5.4467 0.0601 90.6953 0

Dummy variable for ownership of air conditioner 0.0644 0.0232 2.7776 0.0056


Access of electricity from private generator (days per week) 0.0182 0.003 6.0164 0

Dependency ratio –0.2628 0.0436 –6.0242 0




0.2048 0.0406 5.0473 0


Dummy variable for ownership of fridge 0.1347 0.0303 4.4452 0

Household size –0.2168 0.0111 –19.5221 0


Dummy variable for male –0.0984 0.0327 –3.0142 0.0026


incomplete

–0.1203 0.0289 –4.1566 0

Dummy variable for house rented by households 0.0944 0.0369 2.5595 0.0106

Dummy variable for household head is employed in the government/public sector 0.0582 0.0203 2.8614 0.0043


Proportion of household members between ages 7 and 17 –0.0922 0.0472 –1.9521 0.0512




Dummy variable for ownership of water heater 0.1102 0.0264 4.1723 0

Appendices

113

Missan Coe�cient Std. Err. t |Prob|>t

Intercept 7.4427 0.8849 8.4107 0



Average days of electricity from private generator in qhada 0.2281 0.1042 2.1896 0.0287

Average days of electricity from public network in nahiya –0.4565 0.1435 –3.1813 0.0015

Average days of electricity from shared generator in qhada 0.1161 0.0315 3.6899 0.0002




0.3723 0.0453 8.2271 0

Dummy variable for male –0.3094 0.0346 –8.9303 0



Dummy variable for household head is unemployed –0.2696 0.0315 –8.553 0

Dummy variable for household head employed in the private sector –0.1321 0.0256 –5.1631 0



Dummy variable for source of electricity is shared generator 0.1034 0.0346 2.9905 0.0028


Share of working age males employed 0.0865 0.0321 2.695 0.0071




–0.1561 0.0282 –5.534 0


Dummy variable for main source of drinking water is river/canal/creek/wheel, open

well/covered well, pond/lake, spring, kehriz (man built spring) or other

–0.1302 0.035 –3.7244 0.0002


114

Basrah Coe�cient Std. Err. t |Prob|>t

Intercept 4.518 0.0423 106.7698 0







Number of children between ages 0 and 6 = 10 –0.8232 0.1265 –6.5098 0


Number of employed male = 0 0.1234 0.0259 4.762 0

Number of employed male = 1 0.1161 0.0179 6.5011 0


Dummy variable for ownership of generator 0.0838 0.0278 3.0124 0.0026

Dummy variable for housing unit is clay house, bamboo house, or other 0.275 0.1794 1.533 0.1255

Dummy variable for married –0.1014 0.0263 –3.8502 0.0001

Dummy variable for house rented by households –0.0907 0.0228 –3.9863 0.0001



Dummy variable for source of electricity is shared generator 0.0537 0.024 2.2341 0.0256





Appendices

115

Basrah Coe�cient Std. Err. t |Prob|>t


Dummy variable for ownership of water heater 0.0977 0.02 4.8759 0

Interaction of urban area with dummy variable for household head’s education is


0.1117 0.0374 2.9894 0.0028

Interaction of urban area with dummy variable for maximum education in the

household is primary level incomplete

–0.1752 0.0269 –6.5183 0

(continued)


116

IHSES IPMM

MeanStd. Err.

[95% Conf. Interval] Mean

Std. Err.

[95% Conf. Interval]

Urban 0.68 0.01 0.66 0.70 0.70 0.00 0.69 0.71

Household size 8.42 0.06 8.29 8.54 7.64 0.02 7.60 7.68

Household size sq. 88.67 1.93 84.89 92.45 71.66 0.44 70.79 72.53

Dependency ratio 0.44 0.00 0.44 0.45 0.44 0.00 0.43 0.44

Number of children between ages 0 and 6 1.84 0.03 1.78 1.89 1.61 0.01 1.59 1.62

Proportion of household members between ages 0 and 6 0.21 0.00 0.20 0.21 0.20 0.00 0.20 0.20

Number of children between ages 7 and 17 2.34 0.03 2.28 2.39 2.14 0.01 2.13 2.16

Proportion of household members between ages 7 and 17 0.27 0.00 0.26 0.27 0.27 0.00 0.27 0.27

Number of household members aged 60 or above 0.40 0.01 0.39 0.42 0.33 0.00 0.32 0.33

Proportion of household members aged 60 or above 0.05 0.00 0.05 0.06 0.05 0.00 0.05 0.05

Number of males employed 1.46 0.01 1.44 1.49 1.29 0.00 1.28 1.30

Share of working age males employed 0.69 0.00 0.69 0.70 0.67 0.00 0.67 0.67

Dummy variable for maximum education in the household is no

education

0.02 0.00 0.02 0.03 0.03 0.00 0.03 0.03

Dummy variable for maximum education in the household is

primary level incomplete

0.09 0.00 0.08 0.10 0.08 0.00 0.08 0.08


primary level complete

0.21 0.00 0.20 0.22 0.19 0.00 0.19 0.19


intermediate level incomplete or complete

0.20 0.00 0.20 0.21 0.21 0.00 0.20 0.21


secondary level incomplete or complete

0.16 0.00 0.15 0.17 0.17 0.00 0.17 0.17

Appendix C: Summary Statistics of Key Variables


Appendices

117

IHSES IPMM

MeanStd. Err.


Std. Err.



institute level complete

0.10 0.00 0.10 0.11 0.10 0.00 0.10 0.10



0.21 0.01 0.20 0.22 0.23 0.00 0.23 0.23

Dummy variable for household head’s age 20–29 0.06 0.00 0.06 0.07 0.07 0.00 0.07 0.07





Dummy variable for household head’s age 70 or more 0.06 0.00 0.05 0.06 0.06 0.00 0.06 0.06

Dummy variable for household head is male 0.90 0.00 0.90 0.91 0.92 0.00 0.92 0.92

Dummy variable for household head is born outside the

governorate

0.13 0.00 0.12 0.13 0.13 0.00 0.12 0.13

Dummy variable for household head is married 0.90 0.00 0.89 0.91 0.91 0.00 0.91 0.91

Dummy variable for household head has no education 0.23 0.00 0.22 0.24 0.23 0.00 0.23 0.23

Dummy variable for household head’s education is primary level

incomplete

0.15 0.00 0.14 0.16 0.11 0.00 0.11 0.12

Dummy variable for household head’s education is primary level

complete

0.28 0.00 0.28 0.29 0.29 0.00 0.29 0.29

Dummy variable for household head’s education is intermediate

level incomplete or complete

0.11 0.00 0.10 0.11 0.12 0.00 0.12 0.12

Dummy variable for household head’s education is secondary level


0.08 0.00 0.07 0.09 0.09 0.00 0.08 0.09

Dummy variable for household head’s education is institute level

complete

0.07 0.00 0.07 0.08 0.07 0.00 0.07 0.07

(continued)



118

IHSES IPMM

MeanStd. Err.


Std. Err.


Dummy variable for household head’s education is graduate/

technical level or post graduate level complete

0.08 0.00 0.07 0.09 0.09 0.00 0.09 0.09

Dummy variable for household head is unemployed 0.28 0.01 0.27 0.29 0.31 0.00 0.31 0.32

Dummy variable for household head is employed in the

government/public sector

0.29 0.01 0.28 0.30 0.29 0.00 0.29 0.30

Dummy variable for household head employed in the private

sector

0.43 0.01 0.42 0.44 0.39 0.00 0.39 0.40

Dummy variable for ownership of cooler 0.89 0.00 0.89 0.90 0.89 0.00 0.89 0.89

Dummy variable for ownership of fridge 0.93 0.00 0.92 0.93 0.94 0.00 0.94 0.95

Dummy variable for ownership of freezer 0.49 0.01 0.48 0.50 0.50 0.00 0.50 0.51

Dummy variable for ownership of washing machine 0.70 0.01 0.68 0.71 0.71 0.00 0.70 0.71

Dummy variable for ownership of generator 0.36 0.01 0.35 0.37 0.28 0.00 0.28 0.29

Dummy variable for ownership of water heater 0.89 0.00 0.88 0.90 0.85 0.00 0.85 0.85

Dummy variable for ownership of air conditioner 0.42 0.01 0.40 0.43 0.40 0.00 0.39 0.41

Dummy variable for ownership of vacuum cleaner 0.30 0.01 0.29 0.31 0.29 0.00 0.28 0.30

Dummy variable for ownership of car 0.32 0.01 0.31 0.33 0.40 0.00 0.40 0.41

Dummy variable for ownership of personal computer 0.21 0.01 0.20 0.22 0.08 0.00 0.08 0.08

Dummy variable for housing unit is house or �at 0.94 0.00 0.93 0.94 0.95 0.00 0.94 0.95

Dummy variable for housing unit is clay house, bamboo house, or

other

0.06 0.00 0.06 0.07 0.05 0.00 0.05 0.06

Dummy variable for principal material of wall is brick 0.45 0.01 0.44 0.46 0.45 0.00 0.44 0.46

Dummy variable for principal material of wall is stone or thermo

stone

0.07 0.00 0.06 0.07 0.05 0.00 0.05 0.06

(continued)


Appendices

119

IHSES IPMM

MeanStd. Err.


Std. Err.


Dummy variable for principal material of wall is cement block,

concrete readymade, or precast

0.43 0.01 0.41 0.44 0.43 0.00 0.42 0.44

Dummy variable for principal material of wall is clay, bamboo, or

other

0.06 0.00 0.05 0.06 0.07 0.00 0.07 0.07

Dummy variable for dwelling owned by household 0.71 0.01 0.69 0.72 0.74 0.00 0.74 0.75

Dummy variable for dwelling owned by private sector 0.21 0.01 0.19 0.22 0.15 0.00 0.15 0.16

Dummy variable for dwelling owned by government/public sector 0.09 0.01 0.07 0.10 0.10 0.00 0.10 0.11

Dummy variable for house owned by household 0.71 0.01 0.69 0.72 0.74 0.00 0.74 0.75

Dummy variable for house rented by households 0.13 0.01 0.12 0.14 0.12 0.00 0.12 0.13

Dummy variable for house provided by employer, free with and

without arrangement with owner, random housing

0.16 0.01 0.14 0.17 0.13 0.00 0.13 0.14

Dummy variable for source of electricity is shared generator 0.84 0.01 0.83 0.86 0.86 0.00 0.86 0.87

Dummy variable for source of electricity is private generator 0.31 0.01 0.30 0.32 0.25 0.00 0.25 0.26

Access of electricity from public network (days per week) 6.91 0.02 6.88 6.94 6.79 0.01 6.78 6.80

Access of electricity from shared generator (days per week) 5.60 0.06 5.49 5.71 5.91 0.02 5.87 5.94

Access of electricity from private generator (days per week) 1.17 0.03 1.10 1.24 0.83 0.01 0.80 0.85

Dummy variable for public sewage network 0.33 0.01 0.31 0.34 0.34 0.01 0.33 0.35

Dummy variable for septic tank 0.51 0.01 0.50 0.53 0.48 0.00 0.47 0.49

Dummy variable for closed drain, open drain, or other 0.16 0.01 0.15 0.17 0.18 0.00 0.17 0.18

Dummy variable for main source of drinking water is public

network

0.88 0.01 0.87 0.90 0.88 0.00 0.88 0.89

Dummy variable for main source of drinking water is river/canal/

creek/wheel, open well/covered well, pond/lake, spring, kehriz

(man built spring) or other

0.09 0.01 0.08 0.10 0.09 0.00 0.09 0.09

(continued)

where are iraq’s poor: mapping poverty in iraq - world...

Documents